Hi. Thanks for the great work!
I would like to know how you encode the bounding boxes. Apparently according to class Detect in lib/models/mot/cstrack.py (first image), you use some sort of square encoding scheme which I couldn't understand. I cannot find any place where you use the encoding scheme of either YOLO (second image) or of RPN from Faster R-CNN (third image).
Here is the function that computes the encoding but you never use this function.
Hi. Thanks for the great work! I would like to know how you encode the bounding boxes. Apparently according to class Detect in lib/models/mot/cstrack.py (first image), you use some sort of square encoding scheme which I couldn't understand. I cannot find any place where you use the encoding scheme of either YOLO (second image) or of RPN from Faster R-CNN (third image).
Here is the function that computes the encoding but you never use this function.