Superlee
@ruizengalways I also wonder about this
@philferriere Any progress? If @alisaaalehi is right, the original implement would have some mistakes.
@alisaaalehi Hi, your comment is not consistent with [L153](https://github.com/alisaaalehi/image_warping/blob/master/modified_tensorflow_warping.py#L153) in your project. Your experiments show that the the flow vectors are strored at source location. However you answer here indiactes...
@HuangLian126 I also meet this problem, do you solve it?
@ranwek @ahmedtalbi @smileflank @eliethesaiyan @satya2550 I also meet this problem, do you fix it?
@Tete-Xiao @CharlesShang @santisy @1292765944 In FPN, I just find "P2, P3, P4, P5", but in the original paper, there is "P6". Where is "P6" in this project? Do I miss...
@LeoCHANGcn Not yet
@bauerdavid Sure you can train on your dataset withour mask. You just need to remove the mask loss and modify the dataloader.
[L58](https://github.com/Superlee506/Mask_RCNN_Humanpose/blob/master/config.py#L58) in the config.
As far as I know, you just need to modify this place. However you also need to implement your own dataloader.