基于深度卷积网络的目标检测技术综述

A Review of Object Detection Technology Based on Deep Convolution Network

目前,基于计算机视觉分析的目标检测技术已被广泛研究并应用在众多学科领域中。本文从卷积神经网络结构(CNN)演化的角度,对基于深度卷积网络的目标检测技术进行分析、比较和总结。首先简要介绍了基于CNN的目标检测技术流程;其次重点分析和比较了以CNN为基础的基于深度卷积网络模型的目标检测技术的发展,针对不同选择的预处理方法进行分类、纵向和横向对比;最后总结了目前研究中存在的问题,并对目标检测技术未来发展进行了展望。

At present, the object detection technology based on computer vision analysis has been widely studied and applied in many fields. In terms of the evolution of modern convolutional neural network（CNN）, this paper analyzes, compares and summarizes the object detection technology based on the deep convolution network. Firstly, the paper introduces the process of object detection based on CNN. Secondly we analyze and compare on the CNN model based on the depth of object detection technology development, according to different selective pretreatment method we have the longitudinal and transverse comparison; finally, the existing problems in the research are summarized and the future development of the object detection technology is prospected.