CoolFace
Apppublic

preller/test-pose

sourceHugging Faceapache-2.0updated 1y agoView on Hugging Face
0likes
App README

MediaPipe Pose Detection API

This Hugging Face Space provides an API for processing videos with MediaPipe's pose detection model. Upload your video and get back a processed version with pose landmarks drawn on it.

API Usage

Process a Video

POST /process-video/

Upload a video file and receive a processed video with pose landmarks.

Example using curl:

bash
curl -X POST -F "file=@your_video.mp4" https://your-space-name.hf.space/process-video/ --output processed_video.mp4

How it Works

The application:

  1. 1.Receives an uploaded video file
  2. 2.Extracts frames from the video
  3. 3.Processes each frame with MediaPipe's pose detection
  4. 4.Draws the pose landmarks on each frame
  5. 5.Combines the processed frames into a new video
  6. 6.Returns the processed video

Technical Details

  • —Uses MediaPipe for efficient pose detection
  • —Built with FastAPI for high-performance API endpoints
  • —Optimized for processing large videos with batch processing and memory management