Automatic pedestrian detection and monitoring system based on Deep Learning

中文文档

Monitoring plays an important role in security and inspections, but it is also a very tedious task. The emergence of deep learning has liberated humans from this task to some extent. This project builds a simple and effective monitoring system based on the goal detection of deep learning, which can automate the flow statistics and pedestrian detection.

This system is based on the Apache2.0 protocol open source, please strictly abide by the open source agreement.

0x00 Introduction

The system consists of the following three sub-projects:

• 1.Pedestrian detection system based on TensorFlow platform
• 2.Push flow system based on Android platform
• 3.JavaWeb-based display system

The overall framework is shown below:

0x01 Server Deployment

1.Server configuration requirements

Configuration	Basic requirements
OS	Ubuntu 16.04 x64
CPU	Main frequency 2.0GHz or more
RAM	8G or more
GPU	NVIDIA GTX1080 or more
Network	The server IP address needs to be the public IP address.

2.Pedestrian detection system based on TensorFlow platform

The system relies on the following:

Dependency	Installation method
Python3.5	Skip
pip	Skip
TensorFlow-1.11.0-GPU	Skip
Python version - OpenCV	Skip
requests	pip3 install requests
frozen_inference_graph.pb	Download Link
Nginx with RTMP	Installation Process

How to run the system：

• Copy the .pb model file obtained after training the model in the pythondirectory；
• Modify the RTMP_HOST variable in the main.pyfile and runmain.py；

3.Push flow system based on Android platform

How to run the system：

• Import the project in the 'android' directory in an integrated development environment such as IDEA or AndroidStudio,and modify the static variavles in 'MainActivity.java'；

4.Display system based on SSM (SpringMVC+Spring+Mybatis) Internet lightweight framework

The system relies on the following

Dependency	Installation method
JDK-1.8.0	Skip
Apache-Tomcat-9.0.12	Skip
Maven	Skip
Mysql	Need to configure remote access rights

How to run the system：

• The system is developed based on the IDEA integrated development environment. The dependencies in the SSM framework are all based on Maven configuration. Import the project under the web directory in Idea, export the war package, and put the war package on the server tomcat/webapps directory, run ./startup.sh to start the tomcat container

0x02 Project Display

• Added visual view of human flow statistics for large data volumes；

• Show the full effect of the pedestrian detection project，Display link;

0x03 About

• How to support the author: Click on star button in the upper right corner is the maximum support of the author；
• If you have questions or discuss the pedestrian detection algorithm model, please submit an issue,thanks;