Open ma3252788 opened 4 years ago
one(former) is to filter some anchors for those cls-score less than threshold , anthor(latter) is to reduce some anchors for overlap
one(former) is to filter some anchors for those cls-score less than threshold , anthor(latter) is to reduce some anchors for overlap
Thanks for the reply, why does the detection result not change after I adjust these two thresholds? I set both to 0.5, but there are still a lot of boxes.
one(former) is to filter some anchors for those cls-score less than threshold , anthor(latter) is to reduce some anchors for overlap
Thanks for the reply, why does the detection result not change after I adjust these two thresholds? I set both to 0.5, but there are still a lot of boxes.
That's because although demo.py takes these arguments, no further filtering is then performed using them. After appending boxes, labels and scores, add this:
dataframe = list(zip(bboxes, labels, bbox_scores))
dataframe = list(filter(lambda x: x[2] > args.threshold*100, dataframe))
bboxes, labels, bbox_scores = list(zip(*dataframe))
Best wishes.
Can you say what these two thresholds represent?
threshold and iou_threshold
Thank you!!!!