This is a multi-panel visualization tool for training and inference of an object detection model (e.g., the YOLO family, SSD, etc). It has the following panels:
- Configuration panel, where you set the paths of different configuration files.
- Training visualization panel, where the model is trained in the background with its loss and accuracy plotted in the frame (left-hand side) and logged in the console (right-hand side).
- Single image detection panel, where an input image and the detection visualization of it are displayed side by side for comparison.
- Multiple image detection panel, where multiple images are detected in a single batch with the results replacing them.
Configuration panel:
Training visualization panel:
Single image detection panel:
Multiple image detection panel:
- Install dependencies with
requirements.txt. - Run
main.py.