SimCC
SimCC copied to clipboard
leeyegy
[ECCV'2022 Oral] PyTorch implementation for: SimCC: a Simple Coordinate Classification Perspective for Human Pose Estimation (http://arxiv.org/abs/2107.03332). Old name: SimDR
epoch[3] It descends  epoch[87] it fluctuating.  but valid_loss looks all right.  ---------- I did not modify the config file.Is this normal?
how to replace the backbone, i want to replace the hrnet to a small backbone
Hi, It's an amazing work! I reproduce the code of SimDR-Tokenpose version based on mmpose, but didn't get a satisfying performance (Actually a very low performance, somewhere goes wrong definitely...)....
https://github.com/leeyegy/SimCC/blob/aa6e089af29718e889680b6226eb7cac107aa21a/README.md?plain=1#L29 `SimDR* (space-aware SimDR)`: SimDR with Laplace or Gaussian label smoothing for spatial relevance of adjacent labels.
loss += self.criterion(coord_x_pred,coord_gt[:,0]).mul(weight).mean() IndexError: too many indices for tensor of dimension 1
For the `Results on higher input resolution Results on the COCO validation set with the input size of 384×288.` Can you share the configure file (as my understand, resnet/simdr/upsample_free for...
