OpenMMLab Pose Estimation Toolbox and Benchmark.
RTMPose series (RTMPose, DWPose, RTMO, RTMW) without mmcv, mmpose, mmdet etc.
Compute 2D human pose and angles from a video or a webcam.
A fully automated, markerless extrinsic camera calibration pipeline using human motion. Features RTMPose or Metrabs, 3D lifting, and real-world metric scaling.
juxtapose: Multi-Person Pose Tracking Inference SDK with RTMDet, YOLOv8, GDino, RTMPose (ONNX) & Trackers (ByteTrack & BotSORT) & Tapnet with custom ROIs + FastAPI GPU exe
Markerless motion capture with GoPro
Quantitative pipeline for soccer player pose estimation in broadcast video: detector selection, pose estimator benchmark and domain adaptation (YOLO26x + RTMPose-X)
Tennis broadcast analysis pipeline: F3ED shot detection + scoreboard OCR + score-delta reconciler that recovers events the model has no class for (aces, double-faults, first-serve faults). Two-phase Colab→local.
C++ library and set of command-line tools for batched keypoint inference on synchronized image and video inputs
Two-stage skeleton-based action recognition: YOLOv8 + RTMPose + ST-GCN, C++ on NVIDIA DeepStream / TensorRT
Browser-only AI library for real-time object detection (YOLOv8n / YOLOv12n / YOLOv26n, MediaPipe EfficientDet-Lite0) and 3D pose estimation (RTMW3D-X COCO-WholeBody 133-keypoint 3D, InstantHMR MHR70 mesh) on ONNX Runtime Web — wasm, WebGL, WebGPU, WebNN.
AI pull-up detection & technique scoring from video — RTMPose pose estimation, rep counting and GTO-standard violation detection. FastAPI CV engine + Flask web platform.
Real-time 3D Hand & Body Pose Estimation with NVIDIA FoundationStereo & RTMPose for Dual-Eye Intelligence Perception (DIP)
Real-time home training posture feedback system using pose estimation and TCN-based exercise quality classification.
Desktop tennis-swing auto-segmentation. Vendored MediaPipe + RTMDet + RTMPose cutting pipeline wrapped as a Python FastAPI backend with a pluggable Electron/CLI/Web UI.
High-Precision Human Velocity Analysis Framework
Warehouse worker activity recognition from CCTV on a CPU laptop: pose estimation (MediaPipe / RTMPose) → small activity classifier → worker re-identification. Two implementations, browser-based labeling and training, honest leave-one-clip-out evaluation. Demo, not a product.
from a gait video to recognition emotion
측면 러닝 영상에서 Halpe-26 자세 좌표와 16개 러닝 피처를 계산하는 기능 검증용 프로토타입
On-device real-time pose estimation with RTMPose + ONNX Runtime. Flutter Web & desktop. Live demo on GitHub Pages.
To associate your repository with the rtmpose topic, visit your repo's landing page and select "manage topics."