Peng Lu
Peng Lu
In ResNeXt blocks, conv2 is a GConv and should be grouped with its child conv3. However, if conv2 is not in any group while we assign a group for conv3,...
## Motivation ## Modification 1. fix a bug about visualizer's dataset_meta. When testing with multiple GPUs, there are multiple processes but only one `visualizer`. The `visualizer.dataset_meta` is manually set to...
## Motivation ## Modification ## BC-breaking (Optional) ## Use cases (Optional) ## Checklist **Before PR**: - [ ] I have read and followed the workflow indicated in the [CONTRIBUTING.md](https://github.com/open-mmlab/mmpose/blob/master/.github/CONTRIBUTING.md) to...
## Motivation ## Modification 1. add CIDHead 2. add a codec for CID named 'DecoupledHeatmap', which means that the heatmaps for each instance are decoupled and represented on different channels...
## Motivation ## Modification 1. modify DEKR encoder 2. modify `BottomupResize` to make the image located at the center of the expanded image 3. update results for HRNet-w32 & HRNet-w48...
## Motivation ## Modification ## BC-breaking (Optional) ## Use cases (Optional) ## Checklist **Before PR**: - [ ] I have read and followed the workflow indicated in the [CONTRIBUTING.md](https://github.com/open-mmlab/mmpose/blob/master/.github/CONTRIBUTING.md) to...
Hi there, thanks for your excellent work on generating diverse and vivid images. I notice that your work uses DWPose for wholebody pose estimation. There is a new SOTA wholebody...
Hi, thanks for releasing this code. May I ask if the settings of training on two photo-edge datasets will be released? And the trained model?
Hello, has anyone successfully converted a PyTorch model of ED-Pose to ONNX? I've attempted the export, but it's progressing very slowly, so I'm uncertain if it will be completed successfully.