Hi, first of all, I appreciate this implementation code provided by longcw. I like your code. I have a small question in faster_rcnn.py from line 255-250 there is a code section about the tp and tf as follows `

for log

    if self.debug:
        maxv, predict = cls_score.data.max(1)
        self.tp = torch.sum(predict[:fg_cnt].eq(label.data[:fg_cnt])) if fg_cnt > 0 else 0
        self.tf = torch.sum(predict[fg_cnt:].eq(label.data[fg_cnt:]))
        self.fg_cnt = fg_cnt
        self.bg_cnt = bg_cnt

`

what I feel confused about is predict[:fg_cnt].eq(label.data[:fg_cnt]). Afer reading the code many times, I didn't see that the first fg_cnt entires in label are defined as the fg label. I also tried to print label.data[:fg_cnt], not all of them always have the fg label. Do I miss something in the code? Thanks.

@wtliao Yeah, I am confused about it,too. Have you figured it out?