Cartucho
Yes, you are right. Do you think you could help me in any of those tasks?
Yeah, and it is indeed the case. It should be using classes; it needs documentation and tests... Currently, I do not have time to do that, but feel free to...
true negatives do not exist in the context of object detection. A True Negative in the case of object detection would correspond to an object that is not in the...
just try out the code, it will show you the three in different colours (:
Yeah, it is there. But I'm not sure if it is correct. Currently, I do not have any sample to test it.
yes, I added that feature so that people were not required to have OpenCV in the beginning. The code should still work without OpenCV.
I forgot to check if cv2 exists in that part of the code. Now it should work even without cv2 installed.
Usually, for each bounding box, we only use the one with highest confidence score
hello, what do you mean by different scale? You mean a different overlap (IoU)?
you mean `small`, `medium` and `large` objects? Like in the YOLOv3 paper?