TryCatch国创
介绍
为提高城市监控管理的效率及信息的利用率，在当今计算机算力大大提升，监控摄像普及的背景下，我们设计了这款可用于智能城市监控管理的反扒辅助工具，其主要功能可分为扒窃，摔倒行为识别报警，多人姿态估计以及基于这些的有效数据统计分析，如自动统计视频中人群流量变化，以及对于异常事件进行自动分析处理和记录。
Introduction
In order to improve the efficiency of urban monitoring and management and the utilization of information, under the current situation of increasing computer computing power and the popularity of surveillance cameras, we designed this anti-rolling aid tool that can be used for intelligent city monitoring and management. Its main functions can be divided into Recognize alarms for plagiarism, multi-person pose estimation and statistical analysis based on these valid data, such as changes in crowd flow in automatic statistics videos, and automatic analysis and recording of abnormal events. Taking into account the powerful self-learning ability of the neural network, it has added the function of fall recognition, which has a wider range of application scenarios in the family.
环境要求
- 硬件
内存: ＞ 8G
显卡: train GTX1080及性能以上的显卡, test 1060及性能以上的显卡
- 软件
操作系统: ubuntu 16.04
nvidia驱动: 396.37
python: 2.0
caffe: 1.0
contect us
mail: 17853137232@163.com
数据
可以通过该链接百度网盘-try_catch_steal_dataset下载，数据已全部开源
展示效果(Display of results)
