verlab / captar-libras-mediapipe

ConvertMediaToPose

This project focuses on running the mediapipe ML model and transforming its output into the original output of the openpose ML model

Run with Docker

• Clone this repository
• Build the Docker image from the repository's root path:

docker build -t <image-name> .

• Create the container from the repository's root path:

docker run -v <frames path on host>:/train_img -v <output path on host>:/output_mediapipe --gpus all --rm -it <image-name>

• Example of use:

docker run -v /homeLocal-projects/captarlibras_finep/analucia/mediapipe/train_img:/train_img -v ./output_mediapipe:/output_mediapipe --gpus all --rm -it captar-libras-mediapipe

• Run in your container:

python3 convert_media_to_pose.py --frames_dir /train_img --output_dir /output_mediapipe