基于深度学习的图像目标检测算法综述

A survey of image object detection algorithm based on deep learning

图像目标检测是找出图像中感兴趣的目标,并确定他们的类别和位置,是当前计算机视觉领域的研究热点。近年来,由于深度学习在图像分类方面的准确度明显提高,基于深度学习的图像目标检测模型逐渐成为主流。首先介绍了图像目标检测模型中常用的卷积神经网络;然后,重点从候选区域、回归和anchor-free方法的角度对现有经典的图像目标检测模型进行综述;最后,根据在公共数据集上的检测结果分析模型的优势和缺点,总结了图像目标检测研究中存在的问题并对未来发展做出展望。

Image object detection is to find out the objects of interest in the image and determine their classifications and locations.It is a research hotspot in the field of computer vision.In recent years,due to the significant improvement in the accuracy of image classification with deep learning,image object detection models based on deep learning have gradually became mainstream.Firstly,the convolutional neural networks commonly used in image object detection were briefly introduced.Then,the existing classical image object detection models were reviewed from the perspective of candidate regions,regression and anchor-free methods.Finally,according to the detection results on the public dataset,the advantages and disadvantages of the models were analyzed,the problems in the image object detection research were summarized and the future development was forecasted.