Motion Key Point Detection AI Competition

[환경]

- google.colab

- pwd- > /content/drive/MyDrive/Dacon/  ( 코랩에서 시작위치)

- 수정 ->  train.csv ( 잘못된 키points들을 수정하여 사용하였습니다)

(최정명님의 글)[https://dacon.io/competitions/official/235701/talkboard/402730?page=1&dtype=recent]

[참조 ]
- https://github.com/HRNet/HigherHRNet-Human-Pose-Estimation ( hrnet)
- https://github.com/leoxiaobin/deep-high-resolution-net.pytorch  (hrnet)
- https://drive.google.com/file/d/1UoJhTtjHNByZSm96W3yFTfU5upJnsKiS/view?usp=sharing (model zoo에서 pretrained된 model 다운로드)
-https://github.com/leoxiaobin/deep-high-resolution-net.pytorch/blob/master/experiments/coco/hrnet/w32_384x288_adam_lr1e-3.yaml ( yaml data 다운)
-  Hrnet의 pretrained 된 모델과 yaml 파일을 다운받아 사용하였습니다.

[Require module]

!pip install timm 
!pip install tqdm 
!pip install -U albumentations
!pip install torch_optimizer
!pip install torchinfo
!pip install detectron2 -f https://dl.fbaipublicfiles.com/detectron2/wheels/cu101/torch1.8/index.html