End-to-End Trainable Multi-Instance Pose Estimation with Transformers

POET

Implementation of POET for pose estimation

Dowload the COCO Dataset and create the folder structure as mentioned below.

+ data 
    + annotations   
        - 1.xml
        _ 2.xml
        .
        .
    + train2017 
        - 1.jpg
        - 2.jpg
        .
        .

Once you have downloaded the dataset, start training ->

python -m torch.distributed.launch --nproc_per_node=<number-of-gpus> --use_env main.py --coco_path ./data/ --batch_size <batch-size>

I trained using 2 Tesla-V100 with a batch size of 6.

python -m torch.distributed.launch --nproc_per_node=2 --use_env main.py --coco_path ./data/ --batch_size 6

python -m torch.distributed.launch --nproc_per_node=2 --use_env main.py --coco_path ./data/ --batch_size <batch-size> --resume ./snapshots/model.pth

python inference.py