Дисертації з теми "YOLOX"
Щоб переглянути інші типи публікацій з цієї теми, перейдіть за посиланням: YOLOX.
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-50 дисертацій для дослідження на тему "YOLOX".
Біля кожної праці в переліку літератури доступна кнопка «Додати до бібліографії». Скористайтеся нею – і ми автоматично оформимо бібліографічне посилання на обрану працю в потрібному вам стилі цитування: APA, MLA, «Гарвард», «Чикаго», «Ванкувер» тощо.
Також ви можете завантажити повний текст наукової публікації у форматі «.pdf» та прочитати онлайн анотацію до роботи, якщо відповідні параметри наявні в метаданих.
Переглядайте дисертації для різних дисциплін та оформлюйте правильно вашу бібліографію.
1
Nikue, Amassah Djahlin. "Analyse vidéo pour la détection, le suivi et la reconnaissance du comportement pour l'animal en situation d'élevage." Electronic Thesis or Diss., Orléans, 2024. http://www.theses.fr/2024ORLE1011.
Анотація:
La reconnaissance d'activités, également connue sous le nom de reconnaissance d'actions, est un domaine de recherche en vision par ordinateur et en apprentissage automatique, avec diverses applications. L'application la plus courante est l'identification et la compréhension des activités humaines à partir de données visuelles, telles que des images ou des vidéos. Les techniques de reconnaissance d'actions peuvent être appliquées également à la surveillance du bétail, où elles contribuent à améliorer le bien-être des animaux, la productivité et les pratiques de gestion agricole. Ainsi, les travaux réalisés dans ce document se situent dans le cadre de l'analyse vidéo pour la détection, le suivi et la reconnaissance du comportement animal en situation d'élevage. Ces travaux sont réalisés au sein de ANIMOV « Animal Movements Observation from Videos », un projet de recherche pluridisciplinaire mis en œuvre sur la période 2019-2023 par un consortium régional en Centre-Val-de-Loire. Ce projet porte principalement sur deux espèces animales : les éléphants et les chèvres. Dans ce mémoire, nos recherches portent sur l'analyse des activités chez les chèvres. Afin de construire notre système d'analyse du comportement, nous avons mis en place un système de détection et de suivi d'objets. Pour la détection nous avons testé et comparé deux méthodes populaires de la littérature : YOLOv4 et Faster R-CNN, sur des bases de données créées par nos soins. Parmi les deux méthodes de détection, YOLOv4 présente de meilleures performances en terme de précision moyenne et est 2.5 fois plus rapide que le Faster R-CNN. Pour le suivi des chèvres, nous avons testé et comparé également deux méthodes populaires de la littérature : SORT et Deep SORT. L'évaluation des deux méthodes de suivi sur les vidéos de test montre une légère amélioration de Deep SORT par rapport à SORT en terme d'association des données. Cependant, SORT reste plus rapide et plus adapté à un système temps réel. Le système de détection et de suivi mis en place, nous permet de réaliser, en temps réel, l'analyse de l'activité générale du troupeau, avec des indicateurs assez proches de la réalité. La principale faiblesse dans notre système est la perte de détection sur certaines images de la vidéo, qui entraîne des échecs dans le suivi. Ainsi, pour améliorer les performances, nous avons proposé une approche qui fusionne les informations des détections précédentes et de l'image courante, dans une nouvelle architecture de détection (YOLOX), afin de mieux détecter tous les objets sans perdre les anciensActivity recognition, also known as action recognition, is a field of research in computer vision and machine learning, with a variety of applications. One of the most common applications is the identification and understanding of human activities from visual data, such as images or videos. Action recognition techniques can also be applied to livestock monitoring, where they can help improve animal welfare, productivity, and farm management practices. Thus, the work conducted in this document falls within the context of video analysis for the detection, monitoring, and recognition of animal behavior in livestock situations. This work is being achieved within ANIMOV "Animal Movements Observation from Videos", a multidisciplinary research project being implemented over the period 2019-2023 by a regional consortium in Centre-Val-de-Loire. This project concerns two main animal species: elephants and goats. In this thesis, our research focuses on activity analysis for goats. We have built an object detection and tracking system to implement our behavior analysis system. For detection, we tested and compared two popular methods from the literature: YOLOv4 and Faster R-CNN, on self-created datasets. Of the two detection methods, YOLOv4 performs better in average accuracy and is 2.5 times faster than Faster R-CNN. For goat tracking, we also tested and compared two popular methods from the literature: SORT and Deep SORT. Evaluation of both tracking methods on test videos shows a slight improvement of Deep SORT over SORT regarding data association. However, SORT is faster and better suited to a real-time system. The detection and tracking system we have set up enables us to analyze the general activity of the livestock in real-time, with indicators that are fairly close to reality. The main limitation of our system is the loss of detection on certain video images, which leads to tracking failures. So, to improve the performance, we proposed an approach that merges information from previous detections and the current image, in a new detection architecture (YOLOX), to better detect all objects without losing the old ones
2
Núñez-Melgar, Espinoza Erika Pamela, Oré Natali Leonor Reyes, Abad Jorge Raúl Salazar, and Vela Anderson Vásquez. "YOLO." Bachelor's thesis, Universidad Peruana de Ciencias Aplicadas (UPC), 2018. http://hdl.handle.net/10757/625370.
Анотація:
El siguiente trabajo de investigación pretende comprobar la viabilidad del proyecto denominado YOLO. Este proyecto propone crear un medio virtual para interrelacionar dos segmentos de intereses o necesidades diferentes: un segmento que desea vender productos y otro segmento que desea obtenerlos participando en un proceso de rifa virtual a precios accesibles.
En la encuesta virtual realizada para conocer el interés del servicio en el mercado, los resultados que se obtuvieron fueron que, un 72% estaría dispuesto a participar en juegos de azar virtuales y que un 52% ha vendido algún producto nuevo o usado por internet.
Para este proyecto se requiere una inversión aproximada de 141,000.00 nuevos soles de los cuales el 60% corresponde a capital de los accionistas y el 40% restante será financiado por una entidad financiera.
El proyecto YOLO que presentamos contiene el plan estratégico, el plan de marketing, el plan de operaciones, recursos humanos, y el plan económico-financiero. En la evaluación de los flujos de caja que genera el proyecto nos indican que el 100% de la inversión inicial se recupera en el primer año. Asimismo, los accionistas, desde el primer año ya tienen libre disposición de efectivo, el cual crece 23% cada año. Por último, el VAN, para ambos flujos, son positivos y la TIR, para ambos flujos, generan tasas mayores al COK de los accionistas.
Concluimos que le proyecto de negocio genera valor a los accionistas, es viable y rentable.The following research work pretends to test out the viability of the project named YOLO. This project proposes to create a virtual setting to interrelate two segments of different interests or needs: a segment that wants to sell products such as technology, clothing or accessories and another segment that wish to get those items participating in a virtual raffle at affordable prices. In a virtual survey conducted to know the interest and value of the service in the market, the results were that 72% of the people would be willing to participate in virtual games of chance and 52% has sold some new or used product by the Internet. Afterward, the proposal was validated through the raffle of a product in which the interest and participation of the users were astounding because it surpassed the expectations. For this project, it is required an investment of approximately S/141,000.00 of which the 60% corresponds to the shareholders capital and the remaining 40% will be financed by a financial institution. Finally, we present for your revision, the assessment process of project YOLO, it contains the strategic plan, the marketing plan, the operations plan, human resources, and the economic-financial plan. For that reason, in this summary more reference has been made to the financial aspect basing us in the evaluation of cash flows that create the project and which indicate that 100% of the initial outlay it is regained during the first year, all the same, happens to the shareholders who have already free disposition of cash on the first year, which grows 23% every year. Additionally, Net Present Value (NPV), for both flows, is positive and the Internal Rate of Return (IRR), for both flows, generates higher rates than the shareholder's opportunity cost of capital It concludes that the business project generates value to the stockholders, it is viable and profitable.
Trabajo de investigación
3
Ferrer, Bustamante Claudia Mariela, Llanos Víctor Hugo Ibarra, and Flores Carlos Rafael Prialé. "Plataforma virtual de Rifa Yolo." Bachelor's thesis, Universidad Peruana de Ciencias Aplicadas (UPC), 2018. http://hdl.handle.net/10757/625450.
Анотація:
El presente proyecto de negocio ha sido trabajado con la finalidad de atender la necesidad de las personas que son usuarias del comercio electrónico ofreciéndoles una manera innovadora de obtener productos por un costo mínimo.
El objetivo de este plan es aproximar productos a los consumidores que tienen el deseo de tenerlos pero que por diversas razones no han podido conseguirlos.
En esta propuesta elaborada para cumplir el deseo de nuestro cliente elegido, se ha trabajado en la identificación de sus principales motivaciones al momento de comprar por internet como son: el ahorro de tiempo, la conveniencia y la búsqueda del mejor precio, lo cual se contrastó con los principales problemas que enfrentan al momento de hacer compras en un espacio físico: prolongada espera para ser atendido y pagar, precios altos de los productos y mala atención. Además, se logró evidenciar sus temores respecto a la experiencia de comprar por internet.
El proyecto de negocio plantea el servicio de sortear productos de marcas reconocidas por nuestro cliente por lo cual se pondrá a la venta boletos de rifas durante un tiempo establecido para cada sorteo, esto se realizará a través de una plataforma virtual ágil y confiable donde el cliente podrá comprar cuantas opciones desee para ganar. Nuestra propuesta de valor propone el envío del producto libre de costo, transmisiones en vivo y notificaciones de todos los sorteos que serán debidamente validados, variedad de productos, garantía, medios de pago sencillos, bonificaciones.This business project has been developed with the purpose of meeting the needs of people who are users of electronic commerce by offering an innovative way to obtain products for a minimal cost. The objective of this plan is to bring products closer to consumers who have the desire to have them but for various reasons have not been able to get them. In this proposal developed to fulfill the desire of our chosen client, we have worked on identifying their main motivations when buying online such as: saving time, convenience and searching for the best price, which was contrasted with the main problems they face when making purchases in a physical space: prolonged waiting to be attended and paid, high prices of products and poor attention. In addition, it was possible to highlight their fears regarding the experience of shopping online. The business project raises the service of raffling products of brands recognized by our client, for which raffle tickets will be sold for a set time for each draw, this will be done through an agile and reliable virtual platform where the customer can buy as many options as he wants in order to win the raffle. Our value proposition proposes sending the product free of cost, live broadcasts and notifications of all the draws that will be duly validated, variety of products, guarantee, simple means of payment, bonuses.
Trabajo de investigación
4
Marmayohan, Nivethan, and Abdirahman Farah. "Scene analysis using Tensorflow & YOLO algorithms on Raspberry pi 4." Thesis, Högskolan i Halmstad, Akademin för informationsteknologi, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:hh:diva-45540.
Анотація:
Objektdetektion är en av de viktigaste mjukvarukomponenterna i nästa generation trafikövervakning. Deep learnings-algoritmer för objektdetektion, exempelvis YOLO (You Only Look Once), är snabba och noggranna algoritmer i realtid. Realtidsdetektion och igenkänning av objekt är viktiga uppgifter för bildbehandling. I denna studie presenteras ett inbäddat system för detektion och igenkänning av objekt i normal videohastighet (realtid). Indata är följaktligen en videoström som härstammar från en trafikmiljö i Halmstad. Hårdvaran är Raspberry pi 4 i vilken programvarupaketen Tensorflow, YOLO samt träningskonceptet ”Transfer learning” har implementerats. Resultaten presenteras i form av kvantifiering av realtidskörning på FPS (frames per second), detektion noggrannhet, CPU-temperatur och CPU-frekvens i olika experiment. En slutsats är att Raspberry pi 4 kan utföra objektklassificering och detektion med hög noggrannhet i en del scenarier för trafikövervakning med YOLO-algoritmer. Ett scenario för att klassificera objekt med långsam hastighet till exempel gående, skulle det vara genomförbar med att klassificera och detektera med en högnoggrannhet. För objekt med höghastighet som bilar och cyklister så har Raspberry pi 4 svårt att detektera och klassificera objekter.Object detection is one of the essential software components in the next generation of traffic monitoring. Real-time detection and recognition of objects are essential tasks for image processing. Therefore, deep learning algorithms for object detection such as YOLO (You Only Look Once) are increasingly used in image analysis, since they run in normal video frame rate (real-time) and are reasonably accurate. This study presents an embedded system and its results for detecting and recognizing objects in real-time. Results are based on a video stream originating from a traffic environment in the city of Halmstad (Sweden). The embedded system is implemented in Raspberry pi 4 using the software Tensorflow and different deep learning algorithms of the YOLO software package. Real-time analyses on frames per second, accuracy in mean average precision, CPU temperature, and CPU frequency are reported for experiments comprising transfer learning. A main conclusion is that Raspberry pi 4 can perform object classification and detection with high accuracy in certain scenarios for traffic monitoring with YOLO algorithms. For example, classifying objects with the speed of a pedestrian would be feasible with classifying and detecting with high accuracy. On the other hand, with high-speed objects such as cars and cyclists, it is a more challenging task for Raspberry pi 4 to detect and classify objects.
5
Al, Hakim Ezeddin. "3D YOLO: End-to-End 3D Object Detection Using Point Clouds." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-234242.
Анотація:
For safe and reliable driving, it is essential that an autonomous vehicle can accurately perceive the surrounding environment. Modern sensor technologies used for perception, such as LiDAR and RADAR, deliver a large set of 3D measurement points known as a point cloud. There is a huge need to interpret the point cloud data to detect other road users, such as vehicles and pedestrians. Many research studies have proposed image-based models for 2D object detection. This thesis takes it a step further and aims to develop a LiDAR-based 3D object detection model that operates in real-time, with emphasis on autonomous driving scenarios. We propose 3D YOLO, an extension of YOLO (You Only Look Once), which is one of the fastest state-of-the-art 2D object detectors for images. The proposed model takes point cloud data as input and outputs 3D bounding boxes with class scores in real-time. Most of the existing 3D object detectors use hand-crafted features, while our model follows the end-to-end learning fashion, which removes manual feature engineering. 3D YOLO pipeline consists of two networks: (a) Feature Learning Network, an artificial neural network that transforms the input point cloud to a new feature space; (b) 3DNet, a novel convolutional neural network architecture based on YOLO that learns the shape description of the objects. Our experiments on the KITTI dataset shows that the 3D YOLO has high accuracy and outperforms the state-of-the-art LiDAR-based models in efficiency. This makes it a suitable candidate for deployment in autonomous vehicles.För att autonoma fordon ska ha en god uppfattning av sin omgivning används moderna sensorer som LiDAR och RADAR. Dessa genererar en stor mängd 3-dimensionella datapunkter som kallas point clouds. Inom utvecklingen av autonoma fordon finns det ett stort behov av att tolka LiDAR-data samt klassificera medtrafikanter. Ett stort antal studier har gjorts om 2D-objektdetektering som analyserar bilder för att upptäcka fordon, men vi är intresserade av 3D-objektdetektering med hjälp av endast LiDAR data. Därför introducerar vi modellen 3D YOLO, som bygger på YOLO (You Only Look Once), som är en av de snabbaste state-of-the-art modellerna inom 2D-objektdetektering för bilder. 3D YOLO tar in ett point cloud och producerar 3D lådor som markerar de olika objekten samt anger objektets kategori. Vi har tränat och evaluerat modellen med den publika träningsdatan KITTI. Våra resultat visar att 3D YOLO är snabbare än dagens state-of-the-art LiDAR-baserade modeller med en hög träffsäkerhet. Detta gör den till en god kandidat för kunna användas av autonoma fordon.
6
Yevsieiev, V., O. Tokarieva, and S. Starikova. "Research of Object Recognition in the Workspace of A Mobile Robot Based on the Yolo Method." Thesis, Кременчуцький національний університет імені Михайла Остроградського, 2022. https://openarchive.nure.ua/handle/document/20421.
Анотація:
One of the hallmarks of the advent of the new industrial revolution, Industry 5.0, is the synergy between autonomous robots and humans. All this is possible with the introduction of collaborative robots into all spheres of human activity.
7
Pini, Mattia. "Sviluppo di un Prototipo di Video Sorveglianza Indoor con Tecnologie YOLO ed OpenCV." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/18178/.
Анотація:
In questa tesi viene effettuato uno studio sulla reale efficienza delle Reti Neurali Artificiali e più in particolare delle Convolutional Neural Network nel settore del riconoscimento immagini e della velocità con cui è possibile ottenere informazioni dettagliate da esse.
Viene presentato quindi un Prototipo di Video Sorveglianza per Ambienti Indoor che sfrutta queste tecnologie in maniera efficiente coniugando lo stato dell'arte per il riconoscimento degli oggetti con algoritmi noti per il riconoscimento facciale.
Con un attento studio si è dimostrato inoltre perchè la rete scelta è risultata ottimale per l'obiettivo della tesi, mostrandone peculiarità e prestazioni.
Vengono inoltre affrontate le problematiche che gli algoritmi di riconoscimento facciale classici possiedono, di come l'inclinazione di un volto possa portare a un non riconoscimento della figura dello stesso, e le soluzioni proposte.
Importante inoltre la scelta del dataset per il riconoscimento facciale, poichè una scelta errata potrebbe portare a falsi positivi.
Il risultato è un sistema che attraverso acquisizione immagini da videocamera e attraverso l'uso di reti neurali riesce a verificare se un determinato oggetto o una persona sono realmente presenti o meno all'interno di un ambiente.
8
Sommer, Ted Robert. "The aquatic ecology of the Yolo Bypass floodplain : evaluation at the species and landscape scales /." For electronic version search Digital dissertations database. Restricted to UC campuses. Access is free to UC campus dissertations, 2002. http://uclibs.org/PID/11984.
9
Головатий, Ігор Богданович, та Ihor Holovatiy. "Комп'ютерна система на основі нейромережі для виявлення зіткнення автомобілів". Bachelor's thesis, Тернопільський національний технічний університет імені Івана Пулюя, 2021. http://elartu.tntu.edu.ua/handle/lib/35429.
Анотація:
Кваліфікаційна робота присвячена розробці системи, що дозволяє визначати серйозні автомобільні зіткнення на відеоряді, записаному камерами дорожнього спостереження. Проведено огляд існуючих систем детектування дорожньо-транспортних пригод. Запропоновано спосіб вирішення проблеми за допомогою нейромережі, Наведено опис алгоритму детектування автокатастроф, здійснено пошук і підготовка вибірки. Проаналізовано алгоритми детектування об'єктів. Для детектування автомобіля на відео вибрано YOLOv3-детектор.
Здійснено порівняння детекторів об'єктів для з'ясування впливу на кінцеву ефективність роботи системи в цілому. Реалізовано алгоритм детектування зіткнення автомобілів в режимі реального часу. Розроблювана система була протестована на реальних даних для визначення зіткнень автомобілів. Отримані практичні результати дозволяють стверджувати про ефективність використання розробки..The qualification work deals with the development of a system that allows you to identify serious car collisions on a video recorded by surveillance cameras. A review of existing road accident detection systems was conducted. The way of the decision of a problem by means of a neural network is offered, the description of algorithm of detection of car accidents is given, search and preparation of sampling is carried out. Object detection algorithms are analyzed. A YOLOv3 detector is selected to detect the car on video. The object detectors are compared to determine the impact on the final efficiency of the system as a whole. The algorithm of car collision detection in real time is implemented. The developed system was tested on real data to determine car collisions. The obtained practical results allow us to assert the effectiveness of the development.
Вступ. 1. Аналіз технічного завдання. 1.1 Огляд систем детектування ДТП. 1.2. Поняття ДТП. 2. Проектна частина. 2.1. Згорткові нейронні мережі. 2.2. Спосіб вирішення проблеми. за допомогою згорткових нейронних мереж. 2.3. Опис алгоритму детектування автокатастроф. 2.4. Пошук і підготовка вибірки. 2.5. Аналіз алгоритмів детектування об'єктів. 3. Практична частина. 3.1. Порівняння детекторів об'єктів для з'ясування впливу на кінцеву ефективність роботи системи в цілому. 3.2. Реалізація алгоритму детектування зіткнення автомобілів в режимі реального часу. 3.3. Отримані результати експериментів. 4. Безпека життєдіяльності, основи хорони праці. Висновки. Список використаних джерел
10
Güven, Jakup. "Investigating techniques for improving accuracy and limiting overfitting for YOLO and real-time object detection on iOS." Thesis, Malmö universitet, Fakulteten för teknik och samhälle (TS), 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:mau:diva-19999.
Анотація:
I detta arbete genomförs utvecklingen av ett realtids objektdetekteringssystem för iOS. För detta ändamål används YOLO, en ett-stegs objektdetekterare och ett s.k. ihoplänkat neuralt nätverk vilket åstadkommer betydligt bättre prestanda än övriga realtidsdetek- terare i termer av hastighet och precision. En dörrdetekterare baserad på YOLO tränas och implementeras i en systemutvecklingsprocess. Maskininlärningsprocessen sammanfat- tas och praxis för att undvika överträning eller “overfitting” samt för att öka precision och hastighet diskuteras och appliceras. Vidare genomförs en rad experiment vilka pekar på att dataaugmentation och inkludering av negativ data i ett dataset medför ökad precision. Hyperparameteroptimisering och kunskapsöverföring pekas även ut som medel för att öka en objektdetekringsmodells prestanda. Författaren lyckas öka modellens mAP, ett sätt att mäta precision för objektdetekterare, från 63.76% till 86.73% utifrån de erfarenheter som dras av experimenten. En modells tendens för överträning utforskas även med resultat som pekar på att träning med över 300 epoker rimligen orsakar en övertränad modell.This paper features the creation of a real time object detection system for mobile iOS using YOLO, a state-of-the-art one stage object detector and convoluted neural network far surpassing other real time object detectors in speed and accuracy. In this process an object detecting model is trained to detect doors. The machine learning process is outlined and practices to combat overfitting and increasing accuracy and speed are discussed. A series of experiments are conducted, the results of which suggests that data augmentation, including negative data in a dataset, hyperparameter optimisation and transfer learning are viable techniques in improving the performance of an object detection model. The author is able to increase mAP, a measurement of accuracy for object detectors, from 63.76% to 86.73% based on the results of experiments. The tendency for overfitting is also explored and results suggest that training beyond 300 epochs is likely to produce an overfitted model.
11
Сторожук, Б. В. "Система розпізнавання об‘єктів на ділянках автошляху за допомогою оглядових камер відеоспостережень". Thesis, Чернігів, 2020. http://ir.stu.cn.ua/123456789/23420.
Анотація:
Сторожук, Б. В. Система розпізнавання об‘єктів на ділянках автошляху за допомогою оглядових камер відеоспостережень : випускна кваліфікаційна робота : 123 "Комп’ютерна інженерія" / Б. В. Сторожук ; керівник роботи Є. В. Нікітенко ; НУ "Чернігівська політехніка", кафедра інформаційних і комп’ютерних систем. – Чернігів, 2020. – 106 с.Об'єктом розробки була інтелектуальна система розпізнавання об‘єктів на ділянках автошляху за допомогою оглядових камер відеоспостережень. Метою кваліфікаційної роботи є створення системи, яка зможе прийняти, обробити та розпізнати об’єкти на зображенні. Під час розробки та проектування, були розглянуті ШНМтрументи для створення системи та їх недоліки. Результати даної роботи можуть бути використанні для отримання статистичних даних по кількості та типу транспорту на ділянці автошляху. Можливе подальше вдосконалення системи шляхом покращення методів розпізнавання об’єктів на зображеннях використовуючи нейронну мережу характерно новітньої архітектури, що дозволяє у свою чергу покращити метод класифікації та підвищити показник точності розпізнавання об‘єктів на ділянках автошляху
The object of the development was an intelligent system for recognition of objects on the road sections with the help of overview cameras of video surveillance. The purpose of qualifying work is to create a system that can receive, process and recognize objects in an image. During development and design, the tools for creating the system and their disadvantages were considered. The results of this work can be used for obtaining statistics on the amount and type of transport on the road section. Possible further Improvement of the system by improving the methods for recognizing objects to be saved using the neural venter is characteristic of the latest architecture, which allows it to improve its classification method in its queue and improve the accuracy of the object recognition on the road sections.
12
Alsalehy, Ahmad, and Ghada Alsayed. "Scenanalys av trafikmiljön." Thesis, Högskolan i Halmstad, Akademin för informationsteknologi, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:hh:diva-44936.
Анотація:
Antalet vägtrafikanter ökar varje år, och med det ökar trängseln. Man har därför gjort undersökningar med hjälp av objektdetektionsalgoritmer på videoströmmar. Genom att analysera data resultat är det möjligt att bygga en bättre infrastruktur, för att minska trafikstockning samt olyckor. Data som analyseras kan till exempel vara att räkna hur många trafikanter som vistas på en viss väg (Slottsbron i Halmstad) under en viss tid. Detta examensarbete undersöker teoretiskt hur en YOLO algoritm samt TensorFlow kan användas för att detektera olika trafikanter. Utvärderingsmetoder som användes i projektet för att få resultatet och dra slutsatser är mAP, träning och testning av egna och andras YOLO modeller samt övervakning av FPS- och temperatur-värden. För att möjliggöra detekteringen av trafikflöde i realtid nyttjades Jetson nano toolkit. Flera olika jämförelser har skapats för att avgöra vilken YOLO modell som är lämpligast. Resultaten från tester av olika YOLO modeller visar att YOLO-TensorFlows implementationer kan detektera trafikanter med en godtagbar noggrannhet. Slutsatsen är att Jetson nano har tillräckligt med processorkraft för att detektera olika trafikanter i realtid med hjälp av original YOLO implementation. Metoderna för att detektera trafikanter är standard och fungerande för analysering av trafikflöden.Testning av mer varierande trafikmiljö under längre tidsperioder krävs för att ytterligare verifiera om Jetson nanos lämplighet.
13
Стахив, Ю. Н., та М. Г. Заворотна. "Классификация объектов в режиме реального времени". Thesis, ХНУРЕ, 2019. http://openarchive.nure.ua/handle/document/8478.
Анотація:
An important and urgent task in machine learning is the introduction and optimization of the technology for classifying objects in real time. For this project, completely local solutions were needed, for none of the existing ones in this area met the requirements of the planned one. Yolo - is an advanced object detection system in real time. It has a wide variety of configurations for any requirements. One of the tasks was the choice of configuration, which we will adapt to meet the objectives of the project. A suitable one was found among them, one that could work quickly even on smartphones or the Raspberry Pi - Tiny YOLO.
14
Mehta, Rinav. "COMPARISON OF REFUSE DECOMPOSITION IN THE PRESENCE AND ABSENCE OF LEACHATE RECIRCULATION AT THE YOLO COUNTY, CALIFORNIA TEST CELLS." NCSU, 2000. http://www.lib.ncsu.edu/theses/available/etd-20001219-110441.
Анотація:
MEHTA, RINAV C. Comparison of Refuse Decomposition in the Presence and Absence of Leachate Recirculation at the Yolo County, California Test Cells. (Under the direction of Morton A. Barlaz.)A side by side comparison of two 8,000-metric ton test cells, one operated with (enhanced) and one without (control) leachate recirculation, was performed to evaluate the effects of leachate recirculation on refuse decomposition at Yolo County, CA. After about three years of operation, refuse was excavated in three borings of the enhanced cell (E1, E2 and E3) and two borings in the control cell (C1 and C2). The objective of this study was to present a comparison of test cell performance with respect to moisture content, settlement, methane production and solids decomposition. Refuse moisture content data show that leachate recirculation resulted in an increase in refuse moisture content, but also show that the refuse in the enhanced cell was not uniformly wet. The average moisture content in E1, E2 and E3 was 38.8, 31.7 and 34.8%, respectively, while the average moisture content in C1 and C2 was 14.6 and 19.2%, respectively. The extent of decomposition was determined by the biochemical methane potential (BMP) and the ratio of cellulose plus hemicellulose to lignin ((C+H)/L). BMP analysis showed the average methane potential in the enhanced and control cells to be 24.0 and 30.9 mL CH4/dry-gm, respectively, and the (C+H)/L of 1.09 and 1.44. These data correlates well with the measured methane production in the enhanced and control cell of 54 and 26 L CH4/wet-kg, respectively. Thus, laboratory and field data shows more decomposition in the enhanced cell relative to the control cell. While the overall averages may not appear significantly different, a closer look at the performance of E1 shows a difference in both moisture content and solids decomposition when compared to the control cell. Hence, the extent of decomposition varies within the enhanced cell. The sampling program conducted for the Yolo County test cells, in concert with data on settlement, methane production and the volume of liquid actually recycled, represents perhaps the most complete set of data available to date on a field-scale leachate recirculation landfill.
15
Sun, Ruiwen. "Detecting Faulty Tape-around Weatherproofing Cables by Computer Vision." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-272108.
Анотація:
More cables will be installed owing to setting up more radio towers when it comes to 5G. However, a large proportion of radio units are constructed high in the open space, which makes it difficult for human technicians to maintain the systems. Under these circumstances, automatic detections of errors among radio cabinets are crucial. Cables and connectors are usually covered with weatherproofing tapes, and one of the most common problems is that the tapes are not closely rounded on the cables and connectors. This makes the tape go out of the cable and look like a waving flag, which may seriously damage the radio systems. The thesis aims at detecting this flagging-tape and addressing the issues. This thesis experiments two methods for object detection, the convolutional neural network as well as the OpenCV and image processing. The former uses YOLO (You Only Look Once) network for training and testing, while in the latter method, the connected component method is applied for the detection of big objects like the cables and line segment detector is responsible for the flagging-tape boundary extraction. Multiple parameters, structurally and functionally unique, were developed to find the most suitable way to meet the requirement. Furthermore, precision and recall are used to evaluate the performance of the system output quality, and in order to improve the requirements, larger experiments were performed using different parameters. The results show that the best way of detecting faulty weatherproofing is with the image processing method by which the recall is 71% and the precision reaches 60%. This method shows better performance than YOLO dealing with flagging-tape detection. The method shows the great potential of this kind of object detection, and a detailed discussion regarding the limitation is also presented in the thesis.Fler kablar kommer att installeras på grund av installation av fler radiotorn när det gäller 5G. En stor del av radioenheterna är dock konstruerade högt i det öppna utrymmet, vilket gör det svårt för mänskliga tekniker att underhålla systemen. Under dessa omständigheter är automatiska upptäckter av fel bland radioskåp avgörande. Kablar och kontakter täcks vanligtvis med väderbeständiga band, och ett av de vanligaste problemen är att banden inte är rundade på kablarna och kontakterna. Detta gör att tejpen går ur kabeln och ser ut som en viftande flagga, vilket allvarligt kan skada radiosystemen. Avhandlingen syftar till att upptäcka detta flaggband och ta itu med frågorna. Den här avhandlingen experimenterar två metoder för objektdetektering, det invändiga neurala nätverket såväl som OpenCV och bildbehandling. Den förstnämnda använder YOLO (You Only Look Once) nätverk för träning och testning, medan i den senare metoden används den anslutna komponentmetoden för detektering av stora föremål som kablarna och linjesegmentdetektorn är ansvarig för utvinning av bandbandgränsen. Flera parametrar, strukturellt och funktionellt unika, utvecklades för att hitta det mest lämpliga sättet att uppfylla kravet. Dessutom används precision och återkallande för att utvärdera prestandan för systemutgångskvaliteten, och för att förbättra kraven utfördes större experiment med olika parametrar. Resultaten visar att det bästa sättet att upptäcka felaktigt väderbeständighet är med bildbehandlingsmetoden genom vilken återkallelsen är 71% och precisionen når 60%. Denna metod visar bättre prestanda än YOLO som hanterar markering av flaggband. Metoden visar den stora potentialen för denna typ av objektdetektering, och en detaljerad diskussion om begränsningen presenteras också i avhandlingen.
16
Матлахов, В. І. "Інтелектуальна система розпізнавання образів у Web-контексті". Master's thesis, Сумський державний університет, 2020. https://essuir.sumdu.edu.ua/handle/123456789/82177.
17
Zeng, Xing. "One Stage Fine- Grained Classification." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-301055.
Анотація:
Fine- grained Visual Classification (FGVC) is a rapidly growing field in image classification. However, it is a challenging task due to subcategories sharing subtle differences. Existing approaches tackle this problem by firstly extracting discriminative regions using part localization or object localization or Region Proposal Networks (RPN), then applying Convolutional Neural Network (CNN) or SVM classifier on those regions. In this work, with the purpose of simplifying the above complicated pipeline while keeping high accuracy, we get inspired by the one- stage object detection model YOLO and design a one- stage end- to- end object detector model for FGVC. Specifically, we apply YOLOv5 as a baseline model and replace its Path Aggregation Network (PANet) structure with Weighted Bidirectional Feature Pyramid Network (BiFPN) structure to efficiently fuse information from different resolutions. We conduct experiments on different classification and localization weight ratios to guide choosing loss weights in different scenarios. We have proved the viability of the one- stage detector model YOLO on FGVC, which has 87.1 % top1 accuracy on the FGVC dataset CUB2002011. Furthermore, we have designed a more accurate one- stage model, achieving 88.1 % accuracy, which is the most accurate method compared to the existing localization state- of- the- art models. Finally, we have shown that the higher the classification loss weight, the faster the convergence speed, while increasing slightly localization loss weight can help achieve a more accurate classification but resulting in slower convergence.Finkornad visuell klassificering (FGVC) är ett snabbt växande fält inom bildklassificering. Det är dock en utmanande uppgift på grund av underkategorier som delar subtila skillnader. Befintliga tillvägagångssätt hanterar detta problem genom att först extrahera diskriminerande regioner med dellokalisering eller objektlokalisering eller Region Proposal Networks (RPN) och sedan tillämpa Convolutional Network eller SVM- klassificering på dessa regioner. I det här arbetet, med syftet att förenkla ovanstående komplicerade rörledning samtidigt som vi håller hög noggrannhet, blir vi inspirerade av enstegs objektdetekteringsmodellen YOLO och designar en enstegs end- to- end objektdetektormodell för FGVC. Specifikt tillämpar vi YOLOv5 som basmodell och ersätter dess Path Aggregation Network (PANet) struktur med en viktad dubbelriktad funktionspyramidnätverk (BiFPN) struktur för att effektivt smälta information från olika upplösningar. Vi utför experiment på olika klassificerings och lokaliseringsviktsförhållanden för att vägleda valet av förlustvikter i olika scenarier. Vi har bevisat livskraften hos enstegsdetektormodellen YOLO på FGVC, som har 87,1 % topp1noggrannhet i FGVC- dataset CUB2002011. Dessutom har vi utformat en mer exakt enstegsmodell som uppnår 88,1 % noggrannhet, vilket är den mest exakta metoden jämfört med befintliga lokaliseringsmodeller. Slutligen har vi visat att ju högre klassificeringsförlustvikten är, desto snabbare är konvergenshastigheten, medan en ökning av lokaliseringsförlustvikten ökar något kan bidra till en mer exakt klassificering men resulterar i långsammare konvergens.
18
Kohmann, Erich. "Tecniche di deep learning per l'object detection." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/19637/.
Анотація:
L’object detection è uno dei principali problemi nell’ambito della computer vision. Negli ultimi anni, con l’avvento delle reti neurali e del deep learning, sono stati fatti notevoli progressi nei metodi per affrontare questo problema.
Questa tesi intende fornire una rassegna dei principali modelli di object detection basati su deep learning, di cui si illustrano le caratteristiche fondamentali e gli elementi che li contraddistinguono dai modelli precedenti.
Dopo un infarinatura iniziale sul deep learning e sulle reti neurali in genere, vengono presentati i modelli caratterizzati da tecniche innovative che hanno portato ad un miglioramento significativo, sia nella precisione e nell’accuratezza delle predizioni, che in termini di consumo di risorse.
Nella seconda parte l’elaborato si concentra su YOLO e sui suoi sviluppi. YOLO è un modello basato su reti neurali convoluzionali, con il quale i problemi di localizzazione e classificazione degli oggetti in un’immagine sono stati trattati per la prima volta come un unico problema di regressione. Questo cambio di prospettiva apportato dagli autori di YOLO ha aperto la strada verso un nuovo approccio all’object detection, facilitando il successivo sviluppo di modelli sempre più precisi e performanti.
19
Norén, Aron. "Enhancing Simulated Sonar Images With CycleGAN for Deep Learning in Autonomous Underwater Vehicles." Thesis, KTH, Matematisk statistik, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-301326.
Анотація:
This thesis addresses the issues of data sparsity in the sonar domain. A data pipeline is set up to generate and enhance sonar data. The possibilities and limitations of using cycleGAN as a tool to enhance simulated sonar images for the purpose of training neural networks for detection and classification is studied. A neural network is trained on the enhanced simulated sonar images and tested on real sonar images to evaluate the quality of these images.The novelty of this work lies in extending previous methods to a more general framework and showing that GAN enhanced simulations work for complex tasks on field data.Using real sonar images to enhance the simulated images, resulted in improved classification compared to a classifier trained on solely simulated images.Denna rapport ämnar undersöka problemet med gles data för djupinlärning i sonardomänen. Ett dataflöde för att generera och höja kvalitén hos simulerad sonardata sätts upp i syfte att skapa en stor uppsättning data för att träna ett neuralt nätverk. Möjligheterna och begränsningarna med att använda cycleGAN för att höja kvalitén hos simulerad sonardata studeras och diskuteras. Ett neuralt nätverk för att upptäcka och klassificera objekt i sonarbilder tränas i syfte att evaluera den förbättrade simulerade sonardatan.Denna rapport bygger vidare på tidigare metoder genom att generalisera dessa och visa att metoden har potential även för komplexa uppgifter baserad på icke trivial data.Genom att träna ett nätverk för klassificering och detektion på simulerade sonarbilder som använder cycleGAN för att höja kvalitén, ökade klassificeringsresultaten markant jämfört med att träna på enbart simulerade bilder.
20
Тарасов, О. Є. "Інтелектуальна система автоматичного керування автомобілем у віртуальній моделі навколишнього середовища". Thesis, Чернігів, 2021. http://ir.stu.cn.ua/123456789/25133.
Анотація:
Тарасов, О. Є. Інтелектуальна система автоматичного керування автомобілем у віртуальній моделі середовища : випускна кваліфікаційна робота : 121 "Інженерія програмного забезпечення" / О. Є. Тарасов ; керівник роботи О. В. Трунова ; НУ "Чернігівська політехніка", кафедра технологій та програмної інженерії. – Чернігів, 2021. – 82 с.Кваліфікаційна робота передбачає дослідження сучасного стану розвитку галузі безпілотних автомобілів, а також основних технологій, які в ній застосовуються; дослідження можливостей використання віртуальних середовищ для перевірки ефективності роботи систем автоматичного керування автомобілем; розробку системи автоматичного керування автомобілем та оцінку ефективності її роботи у віртуальній моделі навколишнього середовища. Віртуальне середовище повинно являти собою модель реального світу зі змінними довколишніми умовами: погодою, ландшафтом, часом доби. При виборі перевага має надаватися тим варіантам, в яких дорожня інфраструктура реалізована більш детально та реалістично. Розроблювана система має складатися з частин, що повинні виконувати наступні функції: - підсистема збору даних: отримання даних з віртуального середовища, їх зберігання у сховищі даних; - підсистема обробки даних: підготовка зібраних даних для реалізації системи автоматичного керування; реалізація обраних технологій на основі зібраних даних; - підсистема керування: інтеграція реалізованої системи автоматичного керування до віртуального середовища.
Qualification work involves the study of the current state of development of the unmanned vehicle industry, as well as the main technologies used in it; research of possibilities of use of virtual environments for check of efficiency of work of systems of automatic control of the car; development of an automatic car control system and evaluation of its efficiency in a virtual model of the environment. The virtual environment should be a model of the real world with changing environmental conditions: weather, landscape, time of day. When choosing, preference should be given to those options in which the road infrastructure is implemented in more detail and realistically. The developed system should consist of parts that must perform the following functions: - data collection subsystem: obtaining data from the virtual environment, storing them in the data warehouse; - data processing subsystem: preparation of collected data for the implementation of automatic control system; implementation of selected technologies based on collected data; - control subsystem: integration of the implemented automatic control system into the virtual environment.
21
Venkatesh, Anirudh. "Object Tracking in Games using Convolutional Neural Networks." DigitalCommons@CalPoly, 2018. https://digitalcommons.calpoly.edu/theses/1845.
Анотація:
Computer vision research has been growing rapidly over the last decade. Recent advancements in the field have been widely used in staple products across various industries. The automotive and medical industries have even pushed cars and equipment into production that use computer vision. However, there seems to be a lack of computer vision research in the game industry. With the advent of e-sports, competitive and casual gaming have reached new heights with regard to players, viewers, and content creators. This has allowed for avenues of research that did not exist prior.
In this thesis, we explore the practicality of object detection as applied in games. We designed a custom convolutional neural network detection model, SmashNet. The model was improved through classification weights generated from pre-training on the Caltech101 dataset with an accuracy of 62.29%. It was then trained on 2296 annotated frames from the competitive 2.5-dimensional fighting game Super Smash Brothers Melee to track coordinate locations of 4 specific characters in real-time. The detection model performs at a 68.25% accuracy across all 4 characters. In addition, as a demonstration of a practical application, we designed KirbyBot, a black-box adaptive bot which performs basic commands reactively based only on the tracked locations of two characters. It also collects very simple data on player habits. KirbyBot runs at a rate of 6-10 fps.
Object detection has several practical applications with regard to games, ranging from better AI design, to collecting data on player habits or game characters for competitive purposes or improvement updates.
22
Khlif, Wafa. "Multi-lingual scene text detection based on convolutional neural networks." Thesis, La Rochelle, 2022. http://www.theses.fr/2022LAROS022.
Анотація:
Cette thèse propose des approches de détection de texte par des techniques d'apprentissage profond pour explorer et récupérer des contenus faiblement structurés dans des images de scène naturelles. Ces travaux proposent, dans un premier temps, une méthode de détection de texte dans des images de scène naturelle basée sur une analyse multi-niveaux des composantes connexes (CC) et l'apprentissage des caractéristiques du texte par un réseau de neurones convolutionnel (CNN), suivie d'un regroupement des zones de texte détectées par une méthode à base de graphes. Les caractéristiques des composantes texte brut/non-texte obtenues à différents niveaux de granularité sont apprises via un CNN. Une deuxième méthode est présentée dans cette thèse inspirée du système YOLO. Le système réalise la détection du texte et l'identification du script simultanément. Nous considérons la tâche de détection de texte multi script comme un problème de détection d'objets, où l'objet est le script du texte. La détection de texte et l'identification des scripts sont réalisées avec une approche holistique en utilisant un réseau neuronal convolutionnel unique. Les évaluations expérimentales de ces approches sont réalisées sur le jeu de données MLT (Multi-Lingual Text dataset), nous avons contribué à la création de ce nouveau jeu de données. Il est composé d'images de scènes naturelles et synthétiques contenant du texte, tels que des panneaux de circulation et publicitaires, des noms de magasins, d'images extraites des réseaux sociaux. Ce type d'images représente l'un des types d'images les plus fréquemment rencontrés sur Internet, à savoir les images avec du texte incorporé dans les réseaux sociauxThis dissertation explores text detection approaches via deep learning techniques towards achieving the goal of mining and retrieval of weakly structured contents in scene images. First, this dissertation presents a method for detecting text in scene images based on multi-level connected component (CC) analysis and learning text component features via convolutional neural networks (CNN), followed by a graph-based grouping of overlapping text boxes. The features of the resulting raw text/non-text components of different granularity levels are learned via a CNN. The second contribution is inspired from YOLO: Real-Time Object Detection system. Both methods perform text detection and script identification simultaneously. The system presents a joint text detection and script identification approach based on casting the multi-script text detection task as an object detection problem, where the object is the script of the text. Joint text detection and script identification strategy is realized in a holistic approach using a single convolutional neural network where the input data is the full image and the outputs are the text bounding boxes and their script. Textual feature extraction and script classification are performed jointly via a CNN. The experimental evaluation of these methods are performed on the Multi-Lingual Text MLT dataset. We contributed in building this new dataset. It is constituted of natural scene images with embedded text, such as street signs and advertisement boards, passing vehicles, user photos in microblog. This kind of images represents one of the mostly encountered image types on the internet which are the images with embedded text in social media
23
Barbazza, Sigfrido. "Deep-learning applicato all'identificazione automatica di frutta in immagini." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2016. http://amslaurea.unibo.it/11526/.
Анотація:
Analisi delle fasi per la realizzazione di uno strumento di supporto gli agricoltori, dalla creazione di un dataset, all'addestramento e test di una rete neurale artificiale, con obiettivo la localizzazione del prodotto agricolo all'interno delle immagini.
24
Acharya, Pradip. "DETECTION AND SEGMENTATION OF DEFECTS IN X-RAY COMPUTED TOMOGRAPHY IMAGE SLICES OF ADDITIVELY MANUFACTURED COMPONENT USING DEEP LEARNING." OpenSIUC, 2021. https://opensiuc.lib.siu.edu/theses/2834.
Анотація:
Additive manufacturing (AM) allows building complex shapes with high accuracy. The X-ray Computed Tomography (XCT) is one of the promising non-destructive evaluation techniques for the evaluation of subsurface defects in an additively manufactured component. Automatic defect detection and segmentation methods can assist part inspection for quality control. However, automatic detection and segmentation of defects in XCT data of AM possess challenges due to contrast, size, and appearance of defects. In this research different deep learning techniques have been applied on publicly available XCT image datasets of additively manufactured cobalt chrome samples produced by the National Institute of Standards and Technology (NIST). To assist the data labeling image processing techniques were applied which are median filtering, auto local thresholding using Bernsen’s algorithm, and contour detection. A convolutional neural network (CNN) based state-of-art object algorithm YOLOv5 was applied for defect detection. Defect segmentation in XCT slices was successfully achieved applying U-Net, a CNN-based network originally developed for biomedical image segmentation. Three different variants of YOLOv5 which are YOLOv5s, YOLOv5m, and YOLOV5l were implemented in this study. YOLOv5s achieved defect detection mean average precision (mAP) of 88.45 % at an intersection over union (IoU) threshold of 0.5. And mAP of 57.78% at IoU threshold 0.5 to 0.95 using YOLOv5M was achieved. Additionally, defect detection recall of 87.65% was achieved using YOLOv5s, whereas a precision of 71.61 % was found using YOLOv5l. YOLOv5 and U-Net show promising results for defect detection and segmentation respectively. Thus, it is found that deep learning techniques can improve the automatic defect detection and segmentation in XCT data of AM.
25
Jurečka, Tomáš. "Detekce a klasifikace létajících objektů." Master's thesis, Vysoké učení technické v Brně. Fakulta elektrotechniky a komunikačních technologií, 2021. http://www.nusl.cz/ntk/nusl-442512.
Анотація:
The thesis deals with the detection and classification of flying objects. The work can be divided into three parts. The first part describes the creation of dataset of flying objects. The reverse image search is used to create the dataset. The next part is a research of algorithms for detection, tracking and classification. Subsequently, the individual algorithms are applied and evaluated. In the last part, the design of hardware components is performed.
26
Valentini, Alice. "Evaluation of deep learning techniques for object detection on embedded systems." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2018. http://amslaurea.unibo.it/15478/.
Анотація:
Area surveying is an important tool used to inspect and study in detail a given area, it is especially useful to monitor the movements and the settlement of populations located in a developing country.
Unmanned Aerial Vehicles (UAV), given the recent developments, could represent a suitable technology in order to carry out this task in an easier and cheaper way.
The use of UAV based surveys techniques poses many challenges in terms of accuracy, speed and efficiency.
The target is to build an autonomous flight system which is able to define optimal flight paths using the gathered information from the environment.
In this thesis we will focus on the development of the perception system which has to capture the desired information with accurate and fast detections.
More in detail, we will explore and evaluate the use of object detection models based on Deep Learning techniques who will sense and collect data which will later use for on-board elaboration. The object detection model has to be accurate in order to detect all the objects encountered on the ground and fast in order to not introduce too much latency into the on-board decision system. Fast and accurate decisions could permit an efficient coverage of the area.
Different embedded platforms will be considered and examined in order to meet the model's computational requirements and to provide an efficient use in terms of battery consumption.
Different training configurations will be tested in order to maximize our detection accuracy metric, minimum average precision (mAP).
The detection speed will be then evaluated on our board using Frame Per Second (FPS) metric. In addition to YOLO we also tested TinyYOLO, a smaller and faster network. Results will be then compared in order to find the best configuration in terms of accuracy/speed.
We will show that our system is able to meet all the requirements even if we do not achieve our ideal detection speed.
27
Кушнір, Іван Ярославович. "Мультисервісна комп’ютерна мережа навчального курсового комбінату". Бакалаврська робота, Хмельницький національний університет, 2021. http://elar.khnu.km.ua/jspui/handle/123456789/10443.
28
Кучерук, Владислав Русланович. "Програмована схема керування світлофорами на базі мікро-ЕОМ для навчального макета, що моделює ситуації на перехресті вулиці". Бакалаврська робота, Хмельницький національний університет, 2021. http://elar.khnu.km.ua/jspui/handle/123456789/10423.
29
Taurone, Francesco. "3D Object Recognition from a Single Image via Patch Detection by a Deep CNN." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/18669/.
Анотація:
This thesis describes the development of a new technique for recognizing the 3D pose of an object via a single image. The whole project is based on a CNN for recognizing patches on the object, that we use for estimating the pose given an a priori model.
The positions of the patches, together with the knowledge of their coordinates in the model, make the estimation of the pose possible through a solution of a PnP problem. The CNN chosen for this project is Yolo.
In order to build the training dataset for the network, a new approach is used. Instead of labeling each individual training image as for the standard supervised learning, the initial coordinates of the patches are propagated on all the other images making use of the pose of the camera for all the pictures.
30
Rehnholm, Jonas. "Battery Pack Part Detection and Disassembly Verification Using Computer Vision." Thesis, Mälardalens högskola, Akademin för innovation, design och teknik, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:mdh:diva-54852.
Анотація:
Developing the greenest battery cell and establishing a European supply of batteries is the main goal for Northvolt. To achieve this, the recycling of batteries is a key enabler towards closing the loop and enabling the future of energy.When it comes to the recycling of electric vehicle battery packs, dismantling is one of of the main process steps.Given the size, weight and high voltage of the battery packs, automatic disassembly using robots is the preferred solution. The work presented in this thesis aims to develop and integrate a vision system able to identify and verify the battery pack dismantling process. To achieve this, two cameras were placed in the robot cell and the object detectors You Only Look Once (YOLO) and template matching were implemented, tested and compared. The results show that YOLO is the best object detector out of the ones implemented. The integration of the vision system with the robot controller was also tested and showed that with the results from the vision system, the robot controller can make informed decisions regarding the disassembly.
31
Donini, Massimo. "Algoritmi di stitching per il rilevamento dell'occupazione di aule in un contesto smart campus." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/19063/.
Анотація:
L'obiettivo di questo progetto è studiare un sistema con dispositivi a basso costo in grado di determinare l'occupazione corrente di un'aula, pensato principalmente per ottimizzare la gestione degli spazi universitari o di grandi edifici, implementandone un prototipo.
A supporto di ciò verranno presentati algoritmi di image stitching per fornire una visione ottimale anche degli ambienti più ampi.
In questo progetto è stato implementato un algoritmo di rilevamento dell'occupazione utilizzando componenti hardware economici che seguono una logica programmata appositamente. L'approccio impiegato si basa sull'utilizzo di camere in posizione prospettica, affidando ad una rete neurale l'elaborazione delle immagini. La motivazione di questa scelta deriva dal fatto che le tecnologie basate su camere sono attualmente più accurate ed utilizzarle in posizione prospettica si è rivelato più adeguato, sulla base della conformazione delle aule con più ingressi.
La rete neurale utilizzata è stata YOLOv3, estremamente veloce ed accurata, mentre l'hardware che è stato scelto dopo averlo confrontato con altri simili è il Raspberry Pi 2 Model B, uno dei single-board computer più venduti al mondo. Questa scelta è stata motivata dalla capacità computazionale richiesta e dall'obiettivo di limitare i costi.
Per poter applicare l'algoritmo di occupancy detection anche a stanze più ampie è stato necessario adottare un metodo di image stitching, col quale si è potuto combinare immagini provenienti da più camere per produrre una foto panoramica. In questo caso si è preferito sviluppare un metodo ad hoc invece di utilizzare software già presenti sul mercato in quanto essi necessitano di una capacità computazionale e di risorse elevate.
32
Scalamandrè, Davide. "Sistema di visione per le gestione automatica dei posti in un parcheggio." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2018.
Анотація:
Questo elaborato si propone di risolvere il problema della ricerca di un parcheggio. Sono tante le grandi città che oggi investono in tecnologia, nel tentativo di fornire servizi intelligenti ai propri cittadini. Questa pratica si è notevolmente sviluppata negli ultimi anni, tanto che è stato coniato il termine Smart City. Spesso, in alcune aree urbane ed in determinati orari, il cittadino per trovare un parcheggio libero impiega molto tempo ed è costretto ad effettuare una ricerca nelle zone limitrofe, sperando che questa porti ad un esito positivo, nel più breve tempo possibile. Ciò non accadrebbe se ci fosse un servizio capace di informare l’utente sull’ubicazione del parcheggio libero più vicino rispetto alla propria posizione. La tecnologia che manca è un sistema di visione robusto per il monitoraggio degli spazi di parcheggio. Si è proceduto, quindi, con una ricerca su come il problema sia stato già affrontato in letteratura. Da tale ricerca è emerso che le attuali tecnologie consentono solo di configurare il sistema e le telecamere in modo da analizzare determinate aree delle immagini prodotte. Ciò consentirebbe soltanto la realizzazione di questo servizio per pochi parcheggi, e non si applicherebbe ad un numero più elevato quale quello relativo alle aree parcheggio di un’intera città. Tali considerazioni ci hanno portato ad individuare una tecnologia capace di localizzare un parcheggio, estrapolando le informazioni da una base di conoscenza. Tutto ciò è realizzabile tramite l’Object Detection. Uno dei sistemi di Object Detection più all’avanguardia è Yolo, un sistema real-time che si occupa del rilevamento di oggetti. Questo sistema utilizza una Convolutional Neural Network per effettuare la previsione di una Bounding Box, che delimita l’oggetto all’interno dell’immagine ed effettua la classificazione dell’oggetto presente all’interno di tale Box. Per poter addestrare questo tipo di sistema è stato utilizzato un Dataset presente in letteratura, PKLot.
33
Melcherson, Tim. "Image Augmentation to Create Lower Quality Images for Training a YOLOv4 Object Detection Model." Thesis, Uppsala universitet, Signaler och system, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-429146.
Анотація:
Research in the Arctic is of ever growing importance, and modern technology is used in news ways to map and understand this very complex region and how it is effected by climate change. Here, animals and vegetation are tightly coupled with their environment in a fragile ecosystem, and when the environment undergo rapid changes it risks damaging these ecosystems severely. Understanding what kind of data that has potential to be used in artificial intelligence, can be of importance as many research stations have data archives from decades of work in the Arctic. In this thesis, a YOLOv4 object detection model has been trained on two classes of images to investigate the performance impacts of disturbances in the training data set. An expanded data set was created by augmenting the initial data to contain various disturbances. A model was successfully trained on the augmented data set and a correlation between worse performance and presence of noise was detected, but changes in saturation and altered colour levels seemed to have less impact than expected. Reducing noise in gathered data is seemingly of greater importance than enhancing images with lacking colour levels. Further investigations with a larger and more thoroughly processed data set is required to gain a clearer picture of the impact of the various disturbances.
34
Giambi, Nico. "Sperimentazione di tecniche di Deep Learning per l'Object Detection." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2020. http://amslaurea.unibo.it/21557/.
Анотація:
Il lavoro svolto in questa tesi ruota intorno alla sperimentazione di tecniche di Deep Learning per l'Object Detection, ovvero la costruzione di un Object Detector a la YOLO partendo da zero testando per ogni parte della costruzione più alternative possibili per verificarne la praticità e correttezza, estrapolando per le varie fasi le soluzioni migliori, sia dal punto di vista funzionale sia per quanto riguarda la semplicità. In questa tesi è stato creato un Object Detector sfruttando MobileNet (una Convolutional Neural Network molto veloce) associata ad un algoritmo in stile YOLO (principalmente YOLOv2) e allenata sul dataset COCO (Common Objects in COntext). Le prove effettuate spaziano in tutti i campi, dalla scelta di usare un modello pre-allenato su un altro dataset alla decisione di alcuni parametri da usare come threshold in fase di post-processing. All'interno della tesi verranno spiegati brevemente i temi principali toccati dall'argomento e tutte le prove svolte, spiegando quali di ognuna di queste sia risultata migliore.
35
Thaung, Ludwig. "Advanced Data Augmentation : With Generative Adversarial Networks and Computer-Aided Design." Thesis, Linköpings universitet, Datorseende, 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-170886.
Анотація:
CNN-based (Convolutional Neural Network) visual object detectors often reach human level of accuracy but need to be trained with large amounts of manually annotated data. Collecting and annotating this data can frequently be time-consuming and financially expensive. Using generative models to augment the data can help minimize the amount of data required and increase detection per-formance. Many state-of-the-art generative models are Generative Adversarial Networks (GANs). This thesis investigates if and how one can utilize image data to generate new data through GANs to train a YOLO-based (You Only Look Once) object detector, and how CAD (Computer-Aided Design) models can aid in this process. In the experiments, different models of GANs are trained and evaluated by visual inspection or with the Fréchet Inception Distance (FID) metric. The data provided by Ericsson Research consists of images of antenna and baseband equipment along with annotations and segmentations. Ericsson Research supplied the YOLO detector, and no modifications are made to this detector. Finally, the YOLO detector is trained on data generated by the chosen model and evaluated by the Average Precision (AP). The results show that the generative models designed in this work can produce RGB images of high quality. However, the quality reduces if binary segmentation masks are to be generated as well. The experiments with CAD input data did not result in images that could be used for the training of the detector. The GAN designed in this work is able to successfully replace objects in images with the style of other objects. The results show that training the YOLO detector with GAN-modified data compared to training with real data leads to the same detection performance. The results also show that the shapes and backgrounds of the antennas contributed more to detection performance than their style and colour.
36
Gustafsson, Simon, and Andreas Persson. "Detecting small and fast objects using image processing techniques : A project study within sport analysis." Thesis, Jönköping University, JTH, Avdelningen för datateknik och informatik, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:hj:diva-54343.
Анотація:
This study has put three different object detecting techniques to the test. The goal was to investigate small and fast-moving objects to see which technique’s performance is most suitable within the sports of Padel. The study aims to cover and explain different affecting conditions that could cause better but also worse performance for small and fast object detection. The three techniques use different approaches for detecting one or multiple objects and could be a guideline for future object detection development. The proposed techniques utilize background histogram calculation, HSV masking with edge detection and DNN frameworks together with the COCO dataset. The process is tested through outdoor video footage across all techniques to generate data, which indicates that Canny edge detection is a prominent suggestion for further research given its high detection rate. However, YOLO shows excellent potential for multiple object detection at a very high confidence grade, which provides reliable and accurate detection of a targeted object. This study’s conclusion is that depending on what the end purpose aims to achieve, Canny and YOLO have potential for future small and fast object detection.
37
Lukáč, Jakub. "Sledování osob v záznamu z dronu." Master's thesis, Vysoké učení technické v Brně. Fakulta informačních technologií, 2020. http://www.nusl.cz/ntk/nusl-417275.
Анотація:
Práca rieši možnosť zaznamenávať pozíciu osôb v zázname z kamery drona a určovať ich polohu. Absolútna pozícia sledovanej osoby je odvodená vzhľadom k pozícii kamery, teda vzhľadom k umiestneniu drona vybaveného príslušnými senzormi. Zistené dáta sú po ich spracovaní vykreslené ako príslušné cesty. Práca si ďalej dáva za cieľ využiť dostupné riešenia čiastkových problémov: detekcia osôb v obraze, identifikácie jednotlivých osôb v čase, určenie vzdialenosti objektu od kamery, spracovanie potrebných senzorových dát. Následne využiť preskúmané metódy a navrhnúť riešenie, ktoré bude v reálnom čase pracovať na uvedenom probléme. Implementačná časť spočíva vo využití akcelerátoru Intel NCS v spojení s Raspberry Pi priamo ako súčasť drona. Výsledný systém je schopný generovať výstup o polohe osôb v zábere kamery a príslušne ho prezentovať.
38
Rexhaj, Kastriot. "Machine visual feedback through CNN detectors : Mobile object detection for industrial application." Thesis, Mittuniversitetet, Institutionen för elektronikkonstruktion, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:miun:diva-36467.
Анотація:
This paper concerns itself with object detection as a possible solution to Valmet’s quest for a visual-feedback system that can help operators and other personnel to more easily interact with their machines and equipment. New advancements in deep learning, specifically CNN models, have been exploring neural networks with detection-capabilities. Object detection has historically been mostly inaccessible to the industry due the complex solutions involving various tricky image processing algorithms. In that regard, deep learning offers a more easily accessible way to create scalable object detection solutions. This study has therefore chosen to review recent literature detailing detection models with a selective focus on factors making them realizable on ARM hardware and in turn mobile devices like phones. An attempt was made to single out the most lightweight and hardware efficient model and implement it as a prototype in order to help Valmet in their decision process around future object detection products. The survey led to the choice of a SSD-MobileNetsV2 detection architecture due to promising characteristics making it suitable for performance-constrained smartphones. This CNN model was implemented on Valmet’s phone of choice, Samsung Galaxy S8, and it successfully achieved object detection functionality. Evaluation shows a mean average precision of 60 % in detecting objects and a 4.7 FPS performance on the chosen phone model. TensorFlow was used for developing, training and evaluating the model. The report concludes with recommending Valmet to pursue solutions built on-top of these kinds of models and further wishes to express an optimistic outlook on this type of technology for the future. Realizing performance of this magnitude on a mid-tier phone using deep learning (which historically is very computationally intensive) sets us up for great strides with this type of technology in the future; and along with better smartphones, great benefits are expected to both industry and consumers.Den här rapporten behandlar objekt detektering som en möjlig lösning på Valmets efterfrågan av ett visuellt återkopplingssystem som kan hjälpa operatörer och annan personal att lättare interagera med maskiner och utrustning. Nya framsteg inom djupinlärning har dem senaste åren möjliggjort framtagande av neurala nätverksarkitekturer med detekteringsförmågor. Då industrisektorn svårare tar till sig högst specialiserade algoritmer och komplexa bildbehandlingsmetoder (som tidigare varit fallet med objekt detektering) så ger djupinlärningsmetoder istället upphov till att skapa självlärande system som är återanpassningsbara och närmast intuitiva i dem fall där sådan teknologi åberopas. Den här studien har därför valt att studera ett par sådana teknologier för att hitta möjliga implementeringar som kan realiseras på något så enkelt som en mobiltelefon. Urvalet har därför bestått i att hitta detekteringsmodeller som är hårdvarumässigt resurssnåla och implementera ett sådant system för att agera prototyp och underlag till Valmets vidare diskussioner kring objekt-detekteringsslösningar. Studien valde att implementera en SSD-MobileNetsV2 modellarkitektur då den uppvisade lovande egenskaper kring hårdvarukraven. Modellen implementerades och utvärderades på Valmets mest förekommande telefon Samsung Galaxy S8 och resultatet visade på en god förmåga för modellen att detektera objekt. Den valda modellen gav 60 % precision på utvärderingsbilderna och lyckades nå 4.7 FPS på den implementerade telefonen. TensorFlow användes för programmering och som stödjande mjukvaruverktyg för träning, utvärdering samt vidare implementering. Studien påpekar optimistiska förväntningar av denna typ av teknologi; kombinerat med bättre smarttelefoner i framtiden kan det leda till revolutionerande lösningar för både industri och konsumenter.
39
Lamberti, Lorenzo. "A deep learning solution for industrial OCR applications." Master's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/19777/.
Анотація:
This thesis describes a project developed throughout a six months internship in the Machine Vision Laboratory of Datalogic based in Pasadena, California.
The project aims to develop a deep learning system as a possible solution for industrial optical character recognition applications. In particular, the focus falls on a specific algorithm called You Only Look Once (YOLO), which is a general-purpose object detector based on convolutional neural networks that currently offers state-of-the-art performances in terms of trade-off between speed and accuracy. This algorithm is indeed well known for reaching impressive processing speeds, but its intrinsic structure makes it struggle in detecting small objects clustered together, which unfortunately matches our scenario: we are trying to read alphanumerical codes by detecting each single character and then reconstructing the final string. The final goal of this thesis is to overcome this drawback and push the accuracy performances of a general object detector convolutional neural network to its limits, in order to meet the demanding requirements of industrial OCR applications. To accomplish this, first YOLO's unique detecting approach was mastered in its original framework called Darknet, written in C and CUDA, then all the code was translated into Python programming language for a better flexibility, which also allowed the deployment of a custom architecture. Four different datasets with increasing complexity were used as case-studies and the final performances reached were surprising: the accuracy varies between 99.75\% and 99.97\% with a processing time of 15 ms for images $1000\times1000$ big, largely outperforming in speed the current deep learning solution deployed by Datalogic. On the downsides, the training phase usually requires a very large amount of data and time and YOLO also showed some memorization behaviours if not enough variability is given at training time.
40
Hamren, Rasmus. "APPLYING UAVS TO SUPPORT THE SAFETY IN AUTONOMOUS OPERATED OPEN SURFACE MINES." Thesis, Mälardalens högskola, Akademin för innovation, design och teknik, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:mdh:diva-53376.
Анотація:
Unmanned aerial vehicle (UAV) is an expanding interest in numerous industries for various applications. Increasing development of UAVs is happening worldwide, where various sensor attachments and functions are being added. The multi-function UAV can be used within areas where they have not been managed before. Because of their accessibility, cheap purchase, and easy-to-use, they replace expensive systems such as helicopters- and airplane-surveillance. UAV are also being applied into surveillance, combing object detection to video-surveillance and mobility to finding an object from the air without interfering with vehicles or humans ground. In this thesis, we solve the problem of using UAV on autonomous sites, finding an object and critical situation, support autonomous site operators with an extra safety layer from UAVs camera. After finding an object on such a site, uses GPS-coordinates from the UAV to see and place the detected object on the site onto a gridmap, leaving a coordinate-map to the operator to see where the objects are and see if the critical situation can occur. Directly under the object detection, reporting critical situations can be done because of safety-distance-circle leaving warnings if objects come to close to each other. However, the system itself only supports the operator with extra safety and warnings, leaving the operator with the choice of pressing emergency stop or not. Object detection uses You only look once (YOLO) as main object detection Neural Network (NN), mixed with edge-detection for gaining accuracy during bird-eye-views and motion-detection for supporting finding all object that is moving on-site, even if UAV cannot find all the objects on site. Result proofs that the UAV-surveillance on autonomous site is an excellent way to add extra safety on-site if the operator is out of focus or finding objects on-site before startup since the operator can fly the UAV around the site, leaving an extra-safety-layer of finding humans on-site before startup. Also, moving the UAV to a specific position, where extra safety is needed, informing the operator to limit autonomous vehicles speed around that area because of humans operation on site. The use of single object detection limits the effects but gathered object detection methods lead to a promising result while printing those objects onto a global positions system (GPS) map has proposed a new field to study. It leaves the operator with a viewable interface outside of object detection libraries.
41
Lukáč, Jakub. "Sledování osob ve videu z dronu." Master's thesis, Vysoké učení technické v Brně. Fakulta informačních technologií, 2021. http://www.nusl.cz/ntk/nusl-445483.
Анотація:
Práca rieši možnosť zaznamenávať pozíciu osôb v zázname z kamery drona a určovať ich polohu. Absolútna pozícia sledovanej osoby je odvodená vzhľadom k pozícii kamery, teda vzhľadom k umiestneniu drona vybaveného príslušnými senzormi. Zistené dáta sú po ich spracovaní vykreslené ako príslušné cesty v grafe. Práca si ďalej dáva za cieľ využiť dostupné riešenia čiastkových problémov: detekcia osôb v obraze, identifikácia jednotlivých osôb v čase, určenie vzdialenosti objektu od kamery, spracovanie potrebných senzorových dát. Následne využiť preskúmané metódy a navrhnúť riešenie, ktoré bude v reálnom čase pracovať na uvedenom probléme. Implementačná časť spočíva vo využití akcelerátoru Intel NCS v spojení s Raspberry Pi priamo ako súčasť drona. Výsledný systém je schopný generovať výstup o polohe detekovaných osôb v zábere kamery a príslušne ho prezentovať.
42
Baroncini, Gian Marco. "analisi dei principali campi del deep learning e delle loro reti neurali." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2021. http://amslaurea.unibo.it/23368/.
Анотація:
La tesi tratta una panoramica del campo innovativo del Deep Learning, analizzandone i principali campi di applicazione (Computer Vision e Natural Language Processing) per poi approfondirne nei dettagli più tecnici le reti neurali utilizzate.
Si inizia la trattazione di questi modelli con le reti convoluzionali e le LSTM per poi descrivere strutture più elaborate e recenti che sono identificate come stato dell'arte di alcuni task fondamentali dell'intelligenza artificiale.
Si è cercato di fornire in questo modo una visione più ampia delle tipologie di reti esistenti, evidenziando un’analisi contrapposta di reti feed-forward e reti ricorrenti, ponendo al lettore uno spunto di riflessione scaturito dalla dualità di queste due branche dell’apprendimento automatico, diverse nell’approccio, ma con il medesimo successo in ambito scientifico per la risoluzione di problemi che agevolano le nostre vite.
Viene conclusa l’argomentazione con un piccolo sguardo al futuro che ci aspetta e che prepotentemente influirà sul modo di approcciarci e vivere il mondo come lo conosciamo oggi.
43
Rispoli, Luca. "Un approccio deep learning-based per il conteggio di persone tramite videocamere low-cost in un contesto Smart Campus." Bachelor's thesis, Alma Mater Studiorum - Università di Bologna, 2019. http://amslaurea.unibo.it/19567/.
Анотація:
I più recenti progressi tecnologici hanno provocato una rapida evoluzione del settore delle tecnologie cosiddette Smart, che, ad oggi, vengono integrati in un vasto numero di sistemi.
La diffusione di tali tecnologie non si è tuttavia limitata a dispositivi ed apparecchiature informatiche, ma ha coinvolto anche altri settori, come quello edilizio, la quale influenza ha dato vita al concetto di "smart building".
Un edificio intelligente ha lo scopo di offrire ai suoi abitanti un elevato livello di comfort, creando un ecosistema in cui i vari dispositivi elettronici possono operare interagendo tra essi in completa autonomia, ponendo tuttavia una
considerevole attenzione al fine di evitare sprechi e ridurre, quanto più possibile, l'impatto ambientale.
Il campus di Cesena è stato costruito secondo questi principi e rappresenta il contesto all'interno del quale si è voluto sviluppare il seguente progetto: un sistema scalabile e a basso costo il quale scopo è quello di monitorare il livello di utilizzo delle aule attraverso il conteggio delle persone effettuato utilizzando dispositivi embedded a basso costo ed algoritmi di intelligenza
arti�ficiale, tale sistema deve essere in grado di operare in piena autonomia e deve offrire, secondo parametri definiti all'interno dell'elaborato di tesi, un certo grado di affidabilità e attendibilità.
L'obiettivo è stato raggiunto tramite l'utilizzo di telecamere collegate a dei single-board computer sul quale sono stati configurati algoritmi di intelligenza artificiale per il riconoscimento di persone, il sistema dispone inoltre di un'applicazione web server che permette di consultare i conteggi effettuati e di ricevere segnalazioni riguardanti eventuali malfunzionamenti.
44
Nguyen, Van Dinh. "Exploitation de la détection de contours pour la compréhension de texte dans une scène visuelle." Electronic Thesis or Diss., Sorbonne université, 2018. http://www.theses.fr/2018SORUS473.
Анотація:
L'intérêt porté à la détection de contours pour la compréhension de texte dans une scène visuelle a été croissant au cours des dernières années comme en témoignent un grand nombre d'applications telles que les systèmes de reconnaissance de plaque d'immatriculation de voiture, les systèmes de navigation, les voitures autonomes basées sur la reconnaissance des panneaux de signalisation, etc. Dans cette recherche, nous abordons les défis de la conception de systèmes de lecture de texte de scène automatique robustes et fiables. Deux étapes majeures du système, à savoir, la localisation de texte dans une scène et sa reconnaissance, ont été étudiées et de nouveaux algorithmes ont été développés pour y remédier. Nos travaux sont basés sur l'observation qu'indiquer des régions de texte de scène primaire qui ont forte probabilité d'être des textes est un aspect important dans la localisation et la reconnaissance de cette information. Ce facteur peut influencer à la fois la précision et l'efficacité des systèmes de détection et de reconnaissance. Inspirées par les succès des recherche de proposition d'objets dans la détection et la reconnaissance objet général, deux techniques de proposition de texte de scène ont été proposées, à savoir l'approche Text-Edge-Box (TEB) et l'approche Max-Pooling Text Proposal (MPT). Dans le TEB, les fonctionnalités bottom-up proposées, qui sont extraites des cartes binaires de contours de Canny, sont utilisées pour regrouper les contours connectés et leur attribuer un score distinct. Dans la technique MPT, une nouvelle solution de groupement est proposée, qui est inspiré de l'approche Max-Pooling. À la différence des techniques de regroupement existantes, cette solution ne repose sur aucune règle heuristique spécifique liée au texte ni sur aucun seuil pour fournir des décisions de regroupement. Basé sur ces résultats, nous avons conçu un système pour comprendre le texte dans une scène visuelle en intégrant des modèles a l'état de l'art en reconnaissance de texte, où une suppression des faux positifs et une reconnaissance de mot peut être traitée simultanément. De plus, nous avons développé un système assisté de recherche de texte dans une scène en construisant une interface web en complément du système de compréhension de texte. Le système peut être consulté via le lien: dinh.ubismart.org:27790. Des expériences sur diverses bases de données publiques montrent que les techniques proposées surpassent les méthodes les plus modernes de reconnaissance de textes sous différents cadres d'évaluation. Le système complet propose surpasse également d'autres systèmes complets de reconnaissance de texte et a été soumis à une compétition de lecture automatique dans laquelle il a montré sa performance et a atteint la cinquième position dans le classement (Dec-2017): http://rrc.cvc.uab.es/?ch=2&com =evaluation&task=4Scene texts have been attracting increasing interest in recent years as witnessed by a large number of applications such as car licence plate recognition systems, navigation systems, self-driving cars based on traffic sign, and so on. In this research, we tackle challenges of designing robust and reliable automatic scene text reading systems. Two major steps of the system as a scene text localization and a scene text recognition have been studied and novel algorithms have been developed to address them. Our works are based on the observation that providing primary scene text regions which have high probability of being texts is very important for localizing and recognizing texts in scenes. This factor can influence both accuracy and efficiency of detection and recognition systems. Inspired by successes of object proposal researches in general object detection and recognition, two state-of-the-art scene text proposal techniques have been proposed, namely Text-Edge-Box (TEB) and Max-Pooling Text Proposal (MPT). In the TEB, proposed bottom-up features, which are extracted from binary Canny edge maps, are used to group edge connected components into proposals and score them. In the MPT technique, a novel grouping solution is proposed as inspired by the max-pooling idea. Different from existing grouping techniques, it does not rely on any text specific heuristic rules and thresholds for providing grouping decisions. Based on our proposed scene text proposal techniques, we designed an end-to-end scene text reading system by integrating proposals with state-of-the-art scene text recognition models, where a false positive proposals suppression and a word recognition can be processed concurrently. Furthermore, we developed an assisted scene text searching system by building a web-page user interface on top of the proposed end-to-end system. The system can be accessed by any smart device at the link: dinh.ubismart.org:27790. Experiments on various public scene text datasets show that the proposed scene text proposal techniques outperform other state-of-the-art scene text proposals under different evaluation frameworks. The designed end-to-end systems also outperforms other scene-text-proposal based end-to-end systems and are competitive to other systems as presented in the robust reading competition community. It achieves the fifth position in the champion list (Dec-2017): http://rrc.cvc.uab.es/?ch=2&com =evaluation&task=4
45
Grahn, Fredrik, and Kristian Nilsson. "Object Detection in Domain Specific Stereo-Analysed Satellite Images." Thesis, Linköpings universitet, Datorseende, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-159917.
Анотація:
Given satellite images with accompanying pixel classifications and elevation data, we propose different solutions to object detection. The first method uses hierarchical clustering for segmentation and then employs different methods of classification. One of these classification methods used domain knowledge to classify objects while the other used Support Vector Machines. Additionally, a combination of three Support Vector Machines were used in a hierarchical structure which out-performed the regular Support Vector Machine method in most of the evaluation metrics. The second approach is more conventional with different types of Convolutional Neural Networks. A segmentation network was used as well as a few detection networks and different fusions between these. The Convolutional Neural Network approach proved to be the better of the two in terms of precision and recall but the clustering approach was not far behind. This work was done using a relatively small amount of data which potentially could have impacted the results of the Machine Learning models in a negative way.
46
Charvát, Michal. "System for People Detection and Localization Using Thermal Imaging Cameras." Master's thesis, Vysoké učení technické v Brně. Fakulta informačních technologií, 2020. http://www.nusl.cz/ntk/nusl-432478.
Анотація:
V dnešním světě je neustále se zvyšující poptávka po spolehlivých automatizovaných mechanismech pro detekci a lokalizaci osob pro různé účely -- od analýzy pohybu návštěvníků v muzeích přes ovládání chytrých domovů až po hlídání nebezpečných oblastí, jimiž jsou například nástupiště vlakových stanic. Představujeme metodu detekce a lokalizace osob s pomocí nízkonákladových termálních kamer FLIR Lepton 3.5 a malých počítačů Raspberry Pi 3B+. Tento projekt, navazující na předchozí bakalářský projekt "Detekce lidí v místnosti za použití nízkonákladové termální kamery", nově podporuje modelování komplexních scén s polygonálními okraji a více termálními kamerami. V této práci představujeme vylepšenou knihovnu řízení a snímání pro kameru Lepton 3.5, novou techniku detekce lidí používající nejmodernější YOLO (You Only Look Once) detektor objektů v reálném čase, založený na hlubokých neuronových sítích, dále novou automaticky konfigurovatelnou termální jednotku, chráněnou schránkou z 3D tiskárny pro bezpečnou manipulaci, a v neposlední řadě také podrobný návod instalace detekčního systému do nového prostředí a další podpůrné nástroje a vylepšení. Výsledky nového systému demonstrujeme příkladem analýzy pohybu osob v Národním muzeu v Praze.
47
Oškera, Jan. "Detekce dopravních značek a semaforů." Master's thesis, Vysoké učení technické v Brně. Fakulta informačních technologií, 2020. http://www.nusl.cz/ntk/nusl-432850.
Анотація:
The thesis focuses on modern methods of traffic sign detection and traffic lights detection directly in traffic and with use of back analysis. The main subject is convolutional neural networks (CNN). The solution is using convolutional neural networks of YOLO type. The main goal of this thesis is to achieve the greatest possible optimization of speed and accuracy of models. Examines suitable datasets. A number of datasets are used for training and testing. These are composed of real and synthetic data sets. For training and testing, the data were preprocessed using the Yolo mark tool. The training of the model was carried out at a computer center belonging to the virtual organization MetaCentrum VO. Due to the quantifiable evaluation of the detector quality, a program was created statistically and graphically showing its success with use of ROC curve and evaluation protocol COCO. In this thesis I created a model that achieved a success average rate of up to 81 %. The thesis shows the best choice of threshold across versions, sizes and IoU. Extension for mobile phones in TensorFlow Lite and Flutter have also been created.
48
BATMUNKH, BAYARMAGNAI, and 白榮. "Real-Time Logo Detector via YOLO." Thesis, 2019. http://ndltd.ncl.edu.tw/handle/az7abw.
Анотація:
碩士國立勤益科技大學
資訊工程系
107
Brand logo detecting is challenging task due to its diversity in size and shape. There are several researchers studied this field and achieved remarkable results using fast region-based convolutional networks and other proposed methods. Although, previous researches are focused on accuracy of the detection only, but not in fast detection field. In contrast, we concerned about real-time logo detection, which can lead to new application opportunities that can be explored by further exploration. In this paper, we extend the state-of-the-art real-time detection architecture YOLOv2 and YOLOv3, and used the FlickrLogos-47 dataset, which is new version of well-known logo dataset, for training to solve the problem. We trained one network for each method. Experiment results show promising results on chosen dataset in real-time detection. In addition, we report a comparison between the methods in mAP metric, show detailed result of each method’s performance on each logo class detection and highlight opportunities and improvements can be explored in future work.
49
林 and 林家平. "A YOLO-based Traffic Counting System." Thesis, 2018. http://ndltd.ncl.edu.tw/handle/2t8chr.
Анотація:
碩士國立中央大學
資訊工程學系
106
Image recognition can be applied in many applications of Intelligent Transportation System (ITS). Through automated traffic flow counting, the traffic information can be presented effectively for a given area. After the existing image recognition model process the monitoring video, the coordinates of objects in each frame can be easily extracted. The extracted object coordinates are then filtered to obtain the required vehicle coordinates. To achieve the function of vehicle counting, it is necessary to identify the relationship of vehicles in different frames, i.e., whether or not they represent the same vehicle. Although the vehicle counting can be achieved by using the tracking algorithm, a short period of recognition failure may cause wrong tracking, which will lead to incorrect traffic counting. In this thesis, we propose a system that utilizes the YOLO framework for traffic flow counting. The system architecture consists of three blocks, including the Detector that generates the bounding box of vehicles, the Buffer which stores coordinates of vehicles, and the Counter which is responsible for vehicle counting. The proposed system requires only to utilize simple distance calculations to achieve the purpose of vehicle counting. In addition, by adding checkpoints, the system is able to alleviate the consequence of false detection. The videos from different locations and angles are used to verify and analyze the correctness and overall efficiency of the proposed system, and the results indicate that our system achieves high counting accuracy under the environment with sufficient ambient light.
50
Farinha, João Simões. "In-vehicle object detection with YOLO algorithm." Master's thesis, 2018. http://hdl.handle.net/1822/64273.
Анотація:
Dissertação de mestrado em Computer Science EngineeringWith the growing computational power that we have at our disposal and the ever-increasing amount of data available the field of machine learning has given rise to deep learning, a subset of machine learning algorithms that have shown extraordinary results in a variety of applications from natural language processing to computer vision. In the field of computer vision, these algorithms have greatly improved the state-of-the-art accuracy in tasks associated with object recognition such as detection. This thesis makes use of one of these algorithms, specifically the YOLO algorithm, as a basis in the development of a system capable of detecting objects laying inside a car cockpit. To this end a dataset is collected for the purpose of training the YOLO algorithm on this task. A comparative analysis of the detection performance of the YOLOv2 and YOLOv3 architectures is performed.Several experiments are performed by modifying the YOLOv3 architecture to attempt to improve its accuracy. Specifically tests are performed in regards to network size, and the multiple outputs present in this network. Explorative experiments are done in order to test the effect that parallel network might have on detection performance. Lastly tests are done to try to find an optimal learning rate and batch size for our dataset on the new architectures.
Com o crescente poder computacional que temos à nossa disposição e o aumento da quantidade dados a que temos acesso o campo de machine learning deu origem ao deep learning um subconjunto de algoritmos de machine learning que têm demonstrado resultados extraordinários numa variedade de aplicações desde processamento de linguagens naturais a visão por computador. No campo de visão por computador estes algoritmos têm levado a enormes progressos na correção de sistemas de deteção de objetos. Nesta tese usamos um destes algoritmos, especificament o YOLO, como base para desenvolver um sistema capaz de detetar objetos dentro de um carro. Dado isto um dataset é recolhido com o propósito de treinar o algoritmo YOLO nesta tarefa. Uma analise comparativa da correção dos algoritmos YOLOv2 e YOLOv3 ´e realizada. Várias técnicas relacionadas com a modificação da arquitetura YOLOv3 são exploradas para otimizar o sistema para o problema especifico de deteção a bordo de veículos. Especificamente testes são realizados no contexto de tamanho da rede e dos múltiplos outputs presentes nesta rede. Experiencias exploratórias são realizadas de forma a testar o efeito que redes parallelas podem ter na correção dos algoritmos. Por fim testes são feitos para tentar encontrar learning rates e batch sizes apropriados para o nosso dataset nas novas arquiteturas.