This project demonstrates multi-person pose estimation using TensorFlow, MediaPipe, and Streamlit. It allows you to upload images or videos to detect persons and visualize their pose landmarks.
The app is hosted on hugging face and it can be accessed via this link.
- Upload an image (.jpg, .jpeg, .png) or a video file (.mp4, .mov, .avi, .mkv)
- Detect persons in the uploaded media using TensorFlow object detection
- Visualize pose landmarks on detected persons using MediaPipe
- Option to draw bounding boxes around detected persons
- Demo mode to display pose estimation on a pre-defined image