468,117 Members | 1,444 Online
Bytes | Developer Community
New Post

Home Posts Topics Members FAQ

Share your developer knowledge by writing an article on Bytes.

Traffic Light Detector using Opencv and YOLOv3 for BEGINNERS

To start, I'm a complete beginner with Computer Vision and I made this project following multiple tutorials. I wanted to create a simple program that would take any input video (in my case, my dashcam footage) and detect/recognize the traffic lights in real-time. For some people, this would be an easy task and would build the program in hours, not me.

This version takes any footage (video) and checks frame-by-frame if there're any traffic lights. The program saves the video (with detected (or not) traffic lights) to another video. This works on (most) Dutch traffic lights, and some traffic lights from other countries that look like Dutch traffic lights. Why? Because I trained the model myself, using my own dashcam footage. There are hundreds of images containing traffic lights. I labeled them myself and trained them using the free GPU from Google Colab. This training process took around 4 days. The AVG loss is around: 0.11

Link to Github Repo: https://github.com/initdebugs/Beginn...-OpenCV-YOLOv3
Mar 2 '21 #1
0 1135

Post your reply

Sign in to post your reply or Sign up for a free account.

Similar topics

8 posts views Thread by razer | last post: by
By using this site, you agree to our Privacy Policy and Terms of Use.