is there something wrong with hard negative mining?

[KaleidoZhouYN]

the implement of hard negative mining actually decrease the anchor number so when positive number = 0,negative number will be = 0 too

[XiaXuehai]

positive number != 0, because https://github.com/XiaXuehai/faceboxes/blob/master/dataset.py#L131. There is an imbalance between the positive and negative examples if you do not use hard negative mining.

[KaleidoZhouYN]

positive number != 0, because https://github.com/XiaXuehai/faceboxes/blob/master/dataset.py#L131. There is an imbalance between the positive and negative examples if you do not use hard negative mining.

yes,the box in the random crop will no be 0 but the positive anchor will be 0 because of the IOU

[XiaXuehai]

When encoding, we set the max IOU to be the positive anchor.