摘要 物体检测就是在图片中找出物体的位置并判别出该物体具体是什么东西。在图像处理领域占有很重要的一席之地,特别是近年来深度学习技术的迅猛发展,使得以突破传统低效的识别算法,大大提高了物体检测的效率和精确度。本文主要讲解使用Pytorch这个深度学习框架实现车辆和行人的识别并就SSD, R-CNN 和 YOLO等系列算法进行验证和分析并相互比较, 以找出适合实际应用场景的模型。
关键字 车辆识别 行人检测 RCNN YOLO SSD
ABSTRACT Object detection is about finding where the object is and recognize what kind of object it is in the picture. It plays an important role in image processing field, especially with the rapidly developing of deep learning algorithm these years. And because of that, it break through the recognition of tradition which is not efficiency and improve the precise and efficiency of object detection. This paper is about to using pytorch which is a deep learning framework to training a model of vehicle and pedestrian and analyse SSD, R-CNN series algorithm and YOLO series algorithm to use a proper model for the different scene.
INDEX ITEMS vehicle recognition, pedestrian detection, RCNN, YOLO, SSD