P_net only do not use negative dataset about mtcnn HOT 6 OPEN

That's it

from mtcnn.

The fuction 'gt_imdb()' will return negative dataset as long as the imglist file in .data/mtcnn/imglists/ contains negative images

from mtcnn.

Another question: if you send a negative data into the net, since neg_data do not have roi information, how could the net process in 'bbpx_pred Loss Layer'?

from mtcnn.

Negative image's gt_bbox is set to 0,0,0,0, so the gradient can be computed normally. And I block the gradient of negative images in the customOp named 'negativemining'.

from mtcnn.

Really thanks a lot, you really helped me a lot.
In conclusion: with both positive and negative images, we calculate 'cls_prob' and 'bbox_pred' loss, however, do not backward update the bbow_pred weigh for negative images?
Am i right?

from mtcnn.

i couldn't find anything about do not backward update the bbow_pred weigh for negative images,

forward either... T.T

from mtcnn.