1.Could you please put the visualization code on Github? thanks
2.if I use my own dataset, the image must be processed as mean.pth.tar train_idxs.npy vaild_idxs.npy?
Please refer to the 'cv2_visualize_keypoints' function in the visualization code .
Yes, you need to have the train and test split, which are store as train_idxs.npy and vaild_idxs.npy. Regarding the mean.pth.tar, you can compute it with your own training data.
1.Could you please put the visualization code on Github? thanks 2.if I use my own dataset, the image must be processed as mean.pth.tar train_idxs.npy vaild_idxs.npy?