Closed Terizian closed 2 years ago
The bbox outputs of the TensorRT YOLO engines are [x, y, w, h] (all values ranging between 0 and 1).
The _postprocess_yolo() function will take those values and output [xmin, ymin, xmax, ymax] (real coordinates). The source code is here:
Oh that makes a lot of sense! Thank you so much for the clarification and thank you as always for the prompt replies.
I have a question with regards to the updates done on this repository. I have used this repo 2 years ago and I believe that the coordinates of the bounding boxes were formatted as x, y, w, h (just like the darknet annotations format). Has this changed to xmin, ymin, xmax, ymax?