search

YYuanAnyVision / mxnet_mtcnn_face_detection

MTCNN face detection
738 stars 314 forks source link

Extra region after alignment #14

Closed hajbabaeim closed 7 years ago

hajbabaeim commented 7 years ago

Hi.

After using extract_image_chips function, in output img I got some extra region which included none face part. how can I remove this section from output of extract_image_chips function?

best.

hajbabaeim commented 7 years ago

for more explanation: I expect this one: org but output is: chip_0 there is some extra region in right and down of output image (the image shifted to left and upside!).

YYuanAnyVision commented 7 years ago

you can skip this function if you want a tight bbox, simple use the networks output to crop the image. (as the plot shows) extract_image_chips adds affine transform which is useful with large pose.

hajbabaeim commented 7 years ago

Thanks for your reply @pangyupo . is there any way for me to have a tight bbox like network's output with affine transformation together?