北京航空航天大学学报 ›› 2020, Vol. 46 ›› Issue (9): 1778-1785.doi: 10.13700/j.bh.1001-5965.2020.0066

• 论文 • 上一篇    下一篇

一种轻量化的多目标实时检测模型

邱博1, 刘翔1, 石蕴玉1, 尚岩峰2   

  1. 1. 上海工程技术大学 电子电气工程学院, 上海 201620;
    2. 公安部第三研究所 物联网技术研发中心, 上海 200031
  • 收稿日期:2020-03-02 发布日期:2020-09-22
  • 通讯作者: 刘翔 E-mail:xliu@sues.edu.cn
  • 作者简介:邱博 男,硕士研究生。主要研究方向:计算机视觉、深度学习、目标检测;刘翔 男,博士,副教授,硕士生导师。主要研究方向:计算机视觉及人工生命;石蕴玉 女,博士,讲师,硕士生导师。主要研究方向:视频大数据智能分析;尚岩峰 男,博士,副研究员。主要研究方向:模式识别和视频大数据。
  • 基金资助:
    国家重点研发计划(2016YFC0801304);上海市“科技创新行动计划”高新技术领域项目(17511106803)

A lightweight multi-target real-time detection model

QIU Bo1, LIU Xiang1, SHI Yunyu1, SHANG Yanfeng2   

  1. 1. School of Electronic and Electrical Engineering, Shanghai University of Engineering Science, Shanghai 201620, China;
    2. Internet of Things Technology R&D Center, The Third Research Institute of the Ministry of Public Security, Shanghai 200031, China
  • Received:2020-03-02 Published:2020-09-22
  • Supported by:
    National Key R & D Program of China (2016YFC0801304); Shanghai Science and Technology Innovation Action Plan in Hi-tech Field (17511106803)

摘要: 为实现公安监控系统内容分析的精准智能及提高服务实战能力,提出一种轻量化的多目标实时检测算法。首先,基于CenterNet检测网络增加了CBNet的多融合阶梯级联结构,有效地解决了主干网络在日常监控中特征提取能力不足的问题;其次,通过模型剪枝压缩网络减少参数量,加快了监控视频分析速度。本文利用部分COCO数据集和自行采集的现场数据进行训练与测试,并与其他主流检测算法(YOLO、Faster-RCNN、SSD等)进行消融实验。实验结果表明:所提模型在公共安全监控中能有效地做到速度与精度的均衡,并具有较强的普适性。

关键词: 目标检测, 深度学习, 模型压缩, 模型蒸馏, 级联融合

Abstract: For the public security monitoring system, a lightweight multi-target real-time detection algorithm is proposed in order to realize the accurate intelligence of the content analysis and improve the actual service ability. First, the multi-fusion gradient cascade structure of CBNet is added based on CenterNet detection network, which effectively solves the problem of insufficient feature extraction capability of the backbone network in daily monitoring videos. Second, the number of parameters is reduced through the model pruning and compression, which can speed up the analysis speed of monitoring videos. During the experiments, the dataset for training and testing consists of a part of COCO datasets and a number of field data collected by ourselves. The ablation experiments are conducted with other mainstream detection algorithms (YOLO, Faster-RCNN, SSD, etc.). The experimental results show that the presented model can effectively balance the speed and precision in the analysis of monitoring videos for public security and has stronger universality.

Key words: target detection, deep learning, model compression, model distillation, cascade fusion

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发