This side project has been done with Python, OpenCV, Matplotlib and (of course) YOLOV3.
First you need to have Python 3.X (.7 is good). Then you will need to "pip install" OpenCV with this command line:
- pip install opencv-python
Then you will need Numpy, Matplotlib:
- pip install numpy
- pip install matplotlib
Finally you will need the config and the weights file of the pre-trained YOLOV3 models. These files are available here :
https://pjreddie.com/darknet/yolo/
In this project I used YOLOv3-320 because is it way better than YOLOv3-tiny in terms of object detection quality even though the FPS dropped down from 15-20 to barely 1 FPS (with a classic CPU). I highly recommend to use this code with a GPU.