sriramcu / yolov3_traffic_violations

An automated system of traffic monitoring to detect violations using yolov3 with Keras backend. Riders not wearing helmet are detected and tickets are sent automatically using ANPR.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

yolov3_traffic_violations

This application is an automated system of traffic monitoring to detect violations. As of now, helmet violations detection is supported, but in future I am planning to add wrong side driving detection. The program automatically generates traffic violation tickets (or challans) based on registration details added to the database.

Demonstration of the application along with the explanation of the source code is shown in this video.

This project is based on three currently existing git repos:

  1. BlcaKHat/yolov3-Helmet-Detection
  2. qqwweee/keras-yolo3
  3. marcbelmont/deep-license-plate-recognition

Each of the above repos have been modified to suit my application and can be found in their respective directories with the original README and LICENSE. Since the first repo is licensed under GPL v3.0, this repo will also be licensed under GPL v3.0. Refer to LICENSE file for details.

Setup

Follow these instructions:
I) Initial installation and Downloading Large Files (>100 MB)

$ git clone https://github.com/sriramcu/yolov3_traffic_violations
$ cd yolov3_traffic_violations
$ chmod +x setup.sh  
$ ./setup.sh

II) Set up API key for ANPR

Get your ANPR API key from here. Name this file as api_key.txt and place it in the current folder(yolov3_traffic_violations).

Usage

$ python3 yolov3_helmet_detection_gui.py

Above command launches the GUI from where you can run the program.


Main menu of the GUI.


Helmet Detected (green box)


Helmet Not Detected (popup indicates violation, shown in challan below)


Challan (ticket) generated by the program.

That's all for now. But much more is still to come in the future! Stay tuned.

About

An automated system of traffic monitoring to detect violations using yolov3 with Keras backend. Riders not wearing helmet are detected and tickets are sent automatically using ANPR.

License:GNU General Public License v3.0


Languages

Language:Python 87.5%Language:C# 11.3%Language:Java 0.9%Language:Shell 0.3%