Bnenne

Improved metrics logging and added early stopping. Added robot vs reid project differentiation.

Developed a CLI for easier training and testing for robot detection and re-identification models. Integrated W&B for hyperparameter tuning, and automated sweeps.

Updated configurations for training and tracking

Made it more interactive. A triangle now follows the cursor and you can change the size with your scroll wheel.

Using winit and softbuffer to create a window, edit the buffer, and display. Create create simple shaders, draw lines, and draw circles.

I trained a new model with more epochs and a large YOLO model then tested it on both tracking systems

Trained a YOLO model for robot detection using labels from roboflow. Then tested different tracking methods (bytetrack, botsort).