namas191297

namas191297 / cigpose-onnx

Public

Lightweight ONNX Runtime inference for CIGPose - state-of-the-art whole-body pose estimation (67.5 AP).

19
2
100% credibility
Found Mar 17, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A user-friendly tool for detecting and drawing human whole-body poses (including hands, feet, and face) on images, videos, or live webcam feeds with high accuracy in cluttered scenes.

How It Works

1
🔍 Discover Pose Finder

You find a handy tool that spots human body positions, hands, feet, and faces in photos or videos, even in tricky crowded scenes.

2
📥 Set Up Quickly

You install it on your computer with a simple download, ready in moments without complicated steps.

3
📦 Download Smart Files

You grab the ready-made brain files from the project's page to power the pose detection.

4
Pick Your Media
📸
Single Photo

Load a picture of people to mark their poses instantly.

🎥
Video Clip

Process a short video to track poses frame by frame.

📹
Live Camera

Use your webcam to see real-time body positions as you move.

5
✨ Run the Magic

Hit go, and it automatically finds people and draws colorful lines for arms, legs, hands, and more – accurate and smooth.

6
💾 Save Your Results

Get back your photo or video with poses overlaid, ready to share or study.

✅ Pose Analysis Done!

Celebrate having clear visualizations of body poses for fitness tracking, dance review, or fun experiments – easy and reliable.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 19 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is cigpose-onnx?

Cigpose-onnx delivers lightweight ONNX runtime inference for CIGPose, a state-of-the-art whole-body pose estimation model hitting 67.5 AP on COCO-WholeBody. Written in Python, it lets you pip install and run top-down pose detection on images, videos, or webcam feeds via a simple CLI command like `cigpose --model model.onnx --video clip.mp4`. No PyTorch or MMPose needed—just grab pre-exported lightweight ONNX models and go.

Why is it gaining traction?

It stands out as a lightweight ONNX model alternative to heavy frameworks, with embedded metadata for plug-and-play input sizes and a model zoo trading speed for accuracy (71MB realtime vs 230MB max perf). The CLI handles detection, cropping, and visualization out-of-box, plus high/mid/low-level Python APIs for custom pipelines. Apache 2.0 licensing and GPU support make it a drop-in for production inference.

Who should use this?

Computer vision engineers building real-time apps like fitness trackers, AR overlays, or surveillance systems. Robotics devs needing robust occlusion handling in cluttered scenes. Python devs prototyping pose estimation without framework lock-in.

Verdict

Solid for quick SOTA pose trials—excellent docs and CLI make it dead simple to evaluate. With 19 stars and 1.0% credibility score, it's early beta; test thoroughly before prod, but worth starring as a lightweight GitHub gem for ONNX pose workflows. (198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.