OpenMMLab Pose Estimation Toolbox and Benchmark.

A central hub for gathering and showcasing amazing projects that extend OpenMMLab with SAM and other exciting features.

Modular, config-driven pipeline for preprocessing Sign Language datasets with pose and video outputs using MediaPipe, MMPose, and YOLO..

Extended COCO-API

💡💡💡awesome compute vision app in gradio

CL-Detection2023 Challenge Official Repository

RTMO pose estimation with pure ONNX Runtime — tiny class + CLI (webcam/image/video). Models via Releases. No MMPose/MMDeploy.

利用3d hand pose keypoint的手部21关键点模型进行指尖位置识别，pygame用作画面显示，ui互动，钢琴声音播放，让我们可以在桌上弹钢琴

Multi-cow tracking, identification, instant ID learning/unlearning system. Operates on videos or images. Datasets, model weights, and other tools are included. <cattle-recognition, cattle-tracking>

A collection of Python scripts and applications for supporting quantitative behavioral observation using videos.

This is an low-code / no-code application that dramatically reduces the complexity of using pose estimation frameworks to extract pose keypoints/landmarks.

Testing pose estimation models - yolo and mmpose

Blender add-on that converts videos of multiple people into fully‑rigged Rigify animations

AI pull-up detection & technique scoring from video — RTMPose pose estimation, rep counting and GTO-standard violation detection. FastAPI CV engine + Flask web platform.

An app to analyze track cyclists. It uses AI (MediaPipe and MMPose) to check body movements, knee angles, and Center of Mass for the Dutch Olympic Committee (NOC*NSF)

Human pose trajectory prediction using MMPose and deep learning models (LSTM, GRU)

Automated REBA ergonomic risk scoring from images and videos using custom 18-keypoint pose estimation (HRNet-W32 + MotionBERT). Includes the full pipeline: model training, video frame sampling, crowdsourced annotation tooling, SME verification, expert shot curation, and score visualization.

多摄像头人体姿态与步态估计系统 OpenMMLab Pose Estimation Toolbox and Benchmark.

Webcam-based real-time CPR coaching system using MMPose pose estimation + Random Forest classifier to assess elbow form, compression frequency and depth against AHA guidelines — running at 30 FPS on consumer hardware.

Extract 2D/3D pose keypoints from video with a local MMPose engine and store per-frame JSON, skeleton overlays, and a JSONL dataset manifest on Backblaze B2 — a Next.js + FastAPI sample for building pose datasets.