#Pose estimation

The project introduces LitePose, an advanced architecture for multi-person pose estimation tailored for edge devices. By refining HRNet-based models, it adopts a streamlined single-branch design enhancing both efficiency and performance. Methods such as Fusion Deconv Head and Large Kernel Convs are utilized to minimize redundancy and boost model capacity. LitePose achieves higher mAP with a 25% computation increase compared to earlier models and significantly lowers latency on mobile platforms, promoting real-time estimation without compromising performance.

This project provides a system for real-time human head pose estimation with ONNX Runtime and OpenCV. Key steps include: face detection with a bounding box, facial landmark identification using a deep learning model, and pose computation via a PnP algorithm. It runs on Ubuntu 22.04 and requires ONNX Runtime 1.17.1 and OpenCV 4.5.4. The code is compatible with video files and webcams, and includes detailed setup and usage instructions. Licensed under the MIT license, it utilizes publicly available datasets, enhancing both accessibility and practical application.