We currenent release the code and models for:

• Top-down

01/18/2022

[Initial commits]:

1. Models with Top-down.

The followed models and logs can be downloaded on Google Drive: total_models, total_logs.

We also release the models on Baidu Cloud: total_models (lqht), total_logs (j7e2).

• All the models are pretrained on ImageNet-1K without Token Labeling and Layer Scale. Reason can be found in issue #12.

Please refer to get_started for installation and dataset preparation.

1. Download the pretrained models in our repository.

2. Simply run the training scripts in exp as followed:

   bash ./exp/top_down_256x192_global_small/run.sh

   Or you can train other models as follower:

   # single-gpu training
   python tools/train.py <CONFIG_FILE> --cfg-options model.backbone.pretrained_path=<PRETRAIN_MODEL> [other optional arguments]
   
   # multi-gpu training
   tools/dist_train.sh <CONFIG_FILE> <GPU_NUM> --cfg-options model.backbone.pretrained_path=<PRETRAIN_MODEL> [other optional arguments] 

[Note]:

• We use global MHRA during training and set the corresponding hyperparameters in the config.py:

  window: False, # whether use window MHRA
  hybrid: False, # whether use hybrid MHRA

• To avoid out of memory, we can use torch.utils.checkpoint in the config.py:

  use_checkpoint=True, # whether use checkpoint
  checkpoint_num=[0, 0, 2, 0], # index for using checkpoint in every stage

# single-gpu testing
python tools/test.py <CONFIG_FILE> <POSE_CHECKPOINT_FILE> --eval mAP

# multi-gpu testing
tools/dist_test.sh <CONFIG_FILE> <POSE_CHECKPOINT_FILE> <GPU_NUM> --eval mAP

This repository is built based on mmpose and HRT repository.