A real-time multi-person human pose estimation system using TensorFlow MoveNet Multipose (Lightning). Built with OpenCV for video and webcam inference, it detects and visualizes keypoints and skeletal connections with confidence-based filtering, optimized for speed and multi-person scenarios.
python opencv computer-vision deep-learning tensorflow video-processing gpu-acceleration pose-estimation ml-project feature-pyramid-network mobilenetv2 human-pose tensorflow-hub keypoint-detection centernet ai-engineering movenet real-time-inference multiperson-tracking webcam-inference
-
Updated
Jan 2, 2026 - Jupyter Notebook