Machine-Learning-Collection
Machine-Learning-Collection copied to clipboard
YOLOv3 confidence threshold for mAP calculation
Hello, thanks for sharing your work! I have a question on the mAP calculation. Using the same confidence threshold of 0.2 my model somehow can't make it to 75-78%, as you reported.
Whereas in the original implementation, confidence threshold for calculating the mAP on Pascal VOC dataset is 0.005 and for the detection 0.2 or something like that, since on a lower threshold it wouldn't be practical (too many boxes). Was that the same approach or did I miss something?