Qu'est-ce que YOLO ?
YOLO, which stands for “You Only Look Once,” is an advanced vision par ordinateur algorithm designed for real-time détection d'objets. Unlike traditional object detection methods that apply a classifier to various parts of an image, YOLO processes the entire image in a single passage en avant through a réseau neuronal. This unique approach allows it to detect and classify multiple objects in a scene quickly and efficiently.
Comment fonctionne YOLO ?
YOLO divides an input image into a grid and assigns bounding boxes and class probabilities to each grid cell. The algorithm predicts multiple bounding boxes per grid cell, which helps it to localize objects accurately. Each bounding box is associated with a score de confiance that indicates the likelihood of the box containing an object and how well it fits the object.
YOLO utilise un réseau de neurones convolutionnels (CNN) for feature extraction, which enables it to recognize patterns in images effectively. The network architecture has evolved through several versions, with YOLOv3 and YOLOv4 being among the most popular and widely used. These versions have improved accuracy and speed, allowing for better detection of small objects and more complex scenes.
Applications de YOLO
YOLO est utilisé dans diverses applications, y compris la surveillance, véhicules autonomes, robotics, and augmented reality. Its ability to process images in real-time makes it suitable for scenarios where immediate feedback is essential, such as traffic monitoring and security systems.
Avantages et limitations
L'avantage principal de YOLO est sa rapidité, ce qui lui permet de détecter des objets en temps réel, ce qui est crucial pour de nombreuses applications. Cependant, il peut rencontrer des difficultés avec de petits objets dans des scènes complexes et peut parfois produire de faux positifs. Malgré ces limitations, YOLO reste un choix populaire pour les développeurs et les chercheurs dans le domaine de la vision par ordinateur.