Hugging Face's logo

Spaces:
beta3
/
Pose_Estimation_MediaPipe
Sleeping

App Files Community
Pose_Estimation_MediaPipe / README.md
beta3's picture
beta3
Update README.md
abdfcac verified
preview code
|
raw
history blame contribute delete
2.69 kB

A newer version of the Gradio SDK is available: 6.1.0

Upgrade
metadata
title: Pose Estimation MediaPipe
emoji: 🧍
colorFrom: purple
colorTo: gray
sdk: gradio
sdk_version: 5.49.1
app_file: app.py
pinned: false
license: mit
short_description: Pose Estimation App for Human Movement

🎯 Pose Estimation - MediaPipe (CPU Optimized)

A lightweight Gradio web app for performing human pose estimation using [MediaPipe]. Optimized for CPU-only environments.

⚠️ Important Warnings

  • ⚠️ Maximum video size: 5 MB (due to CPU processing limits).
  • ⚠️ Processing time: Long videos may take several minutes to process.

🧠 Description

This project provides an easy-to-use interface to run human pose estimation on videos.
It uses MediaPipe Pose to detect and visualize human body landmarks frame by frame.

You can choose between:

  1. Pose on original video — overlays pose landmarks directly on the original video.
  2. Pose only (black background) — displays only the pose skeleton on a dark background.

⚙️ Parameters

Parameter Description
min_detection_confidence Minimum confidence threshold for the model to detect a pose.
min_tracking_confidence Minimum confidence for the model to track poses across frames.

Both values range from 0.0 to 1.0.
Higher values increase accuracy but may slow down performance on CPU.

🧩 Project Structure 

📂 Projects_Repository/
 ├── pose_estimation_app.py     # Main Gradio app
 ├── requirements.txt           # Python dependencies
 ├── README.md                  # Project documentation

🧑‍💻 Tech Stack

  • MediaPipe Pose – Pose detection and tracking
  • OpenCV – Video processing and frame manipulation
  • NumPy – Array and numerical operations
  • Gradio – Web-based user interface

🧠 Notes

  • The app is CPU-only, optimized for lightweight use.
  • For GPU acceleration, consider using TensorFlow or PyTorch-based pose models.
  • Works best with short videos and well-lit subjects.

🔗 Visit the Repository

👉 GitHub: 1-echo / Projects_Repository

🪪 License

This project is released under the MIT License.
You are free to use, modify, and distribute it with attribution.

🙌 Acknowledgements