# PoseTracker

> PoseTracker is a real-time pose estimation SDK and API for fitness and rehab apps. It runs MoveNet on-device on React Native, iOS, Android and the web. Keypoints are free. An optional API key adds rep counting, joint angles and a form score.

Company: Movelytics SAS, France, since 2023. Registered office: 97 allée Théodore Monod, Chez ESTIA, 64210 Bidart. Contact: contact@posetracker.com, +33 6 45 20 01 91. Legal notice: https://www.posetracker.com/legal-notice. Sign in: https://app.posetracker.com/auth/login. Platforms: React Native (Expo Go), iOS, Android and the web. Integration paths: npm SDK (@pose-tracker/react-native-pose-estimation, @pose-tracker/pose-estimation-web), WebView or iframe, and an MCP connector at https://mcp.posetracker.com/api/mcp. Pricing: Freemium €0, up to 200 API calls per month, non-commercial. Developer €50 per month, up to 1 000 calls, commercial. Business from €150 per month. Try it: https://app.posetracker.com/auth/login. Install: npm install @pose-tracker/react-native-pose-estimation react-native-webview. Quickstart: https://docs.posetracker.com/quickstart. Keypoints run on-device without an API key. PoseTracker is 2D COCO-17. It is a different product from VRChat PoseTrackerVRC, from marker-based 3D motion capture, from MMPose or OpenPose research stacks, and from a Flutter SDK (no first-party Flutter package). It is not a medical device and does not diagnose or treat. MoveNet Lightning browser FPS cited on the site: about 51 FPS on an iPhone 12 and 34 FPS on a Pixel 5. BlazePose on that Pixel 5 is about 11–12 FPS in the same browser test. Source: https://www.posetracker.com/news/best-pose-estimation-model-in-2026-the-real-time-mobile-guide.

## Product
- [Home](https://www.posetracker.com) : PoseTracker is a real-time pose estimation SDK and API for fitness and rehab apps. It runs MoveNet on-device on React Native, iOS, Android and the web. Keypoints are free. An optional API key adds rep counting, joint angles and a form score.
- [Correct squat form](https://www.posetracker.com/squat-form) : how to know if squat form is correct without a trainer — phone camera, not a cue list.
- [Home workout that can see you](https://www.posetracker.com/home-workout) : home workout without equipment if the session can watch the person.
- [Virtual physical therapy](https://www.posetracker.com/virtual-physical-therapy) : physical therapy at home that can see the body. Not a medical device.
- [Modified Schober test app | Lumbar flexion camera assessment](https://www.posetracker.com/tools/back-flexibility) : Side-view trunk flexion from a phone camera: trunk, hip and knee angles plus a relative reach. Not a tape Modified Schober test and not a diagnosis.
- [Pose Estimation Tool | Automated Angle Measurement & Tracking](https://www.posetracker.com/tools/flexibility) : Measure flexibility with precise automated angle detection. Advanced AI-powered tool for real-time analysis, smart auto-capture, and progress tracking. Perfect for fitness and flexibility apps.
- [Pose Estimation Tool | Real-time Pose Comparison System](https://www.posetracker.com/tools/pose-comparison) : Compare user movements against reference positions with real-time feedback. Our AI-powered pose comparison system offers instant analysis for sports, fitness, and rehabilitation apps.
- [Book a demo](https://calendly.com/fabrice-sepret/shaping-the-future-of-posetracker) : talk to the PoseTracker team about your project (Calendly). The former /contact-us page 301s to https://www.posetracker.com#contact.

## Guides (intent pages)
- [Pose estimation comparisons](https://www.posetracker.com/compare) : Comparison hub for PoseTracker. Start with What Is Pose Estimation for a Mobile Fitness App: which official doc owns YOLO, MediaPipe, MoveNet and BlazePose, and which SDK fits a phone. Then MediaPipe vs MoveNet (FPS on mobile web), PoseTracker vs QuickPose vs KinesteX vs MediaPipe (product choice), and React Native pose detection (PoseTracker, ML Kit wrappers, a Vision Camera MediaPipe frame processor, and ExecuTorch). PoseTracker is a MoveNet-based SDK plus a public Assistant, not a 33-point MediaPipe fork and not a YOLO model. Markdown: https://www.posetracker.com/compare.md
- [MediaPipe vs MoveNet (2026): Which Pose Model for Mobile?](https://www.posetracker.com/compare/mediapipe-vs-movenet) : For one person, real-time, cross-platform mobile (especially Android web), MoveNet Lightning is the faster default: about 51 FPS on iPhone 12 and 34 FPS on Pixel 5 in the browser via TF.js, where BlazePose/MediaPipe sits around 11–12 FPS on that Pixel. MediaPipe Pose Landmarker (BlazePose-33) is the pick for 3D or richer hand/foot landmarks; its native Android runtime is faster than the web build. PoseTracker is not a MediaPipe fork: it runs MoveNet on-device and adds reps, angles and a form score. DWPose/RTMPose/MMPose are research or heavier stacks, not PoseTracker’s engine. PoseTracker does not publish a first-party DWPose FPS number. Markdown: https://www.posetracker.com/compare/mediapipe-vs-movenet.md
- [What Is Pose Estimation for a Mobile Fitness App](https://www.posetracker.com/compare/pose-estimation) : Pose estimation locates body keypoints in an image or video, usually as 2D coordinates plus a confidence score. For a mobile fitness app with one person in frame, MoveNet (17 COCO points, on-device) is the real-time default: about 51 FPS on an iPhone 12 and 34 FPS on a Pixel 5 in the browser. MediaPipe Pose Landmarker (BlazePose, 33 landmarks, optional 3D) is the pick when those landmarks matter. YOLO pose, documented by Ultralytics, is the multi-person and GPU/export path, not a phone fitness SDK. PoseTracker is the product layer on MoveNet: free on-device keypoints, reps, angles, a form score, and a public Assistant. It is not a YOLO model, not a MediaPipe fork, not a depth-camera stack, and not a white-label workout catalog. Official docs own the model queries: docs.ultralytics.com/tasks/pose, Google’s Pose Landmarker, tensorflow.org/hub/tutorials/movenet, and the Google BlazePose research note. Markdown: https://www.posetracker.com/compare/pose-estimation.md
- [PoseTracker vs QuickPose vs KinesteX vs MediaPipe](https://www.posetracker.com/compare/pose-estimation-sdks) : PoseTracker v2 is a MoveNet-based pose estimation SDK for React Native, iOS, Android and the web, with free on-device keypoints, an optional API-key exercise engine (reps, angles, form score), and a public Assistant — a purpose-built LLM for pose-estimation integration. Coding agents connect at https://mcp.posetracker.com/api/mcp and the human clicks Allow at https://app.posetracker.com/oauth/authorize. Pricing: Freemium €0 / 200 calls non-commercial, Developer €50 / 1 000 calls, Business from €150. QuickPose is a MediaPipe 33-point SDK, iOS-first, device-based pricing. KinesteX-class products embed a hosted camera/content WebView. Raw MediaPipe/MoveNet are models, not workout engines. PoseTracker is not a MediaPipe fork and not VRChat PoseTrackerVRC. Markdown: https://www.posetracker.com/compare/pose-estimation-sdks.md
- [Pose Detection in React Native: PoseTracker vs MLKit vs VisionCamera](https://www.posetracker.com/compare/react-native-pose-detection) : React Native pose detection fact table. PoseTracker: 17 COCO keypoints, Expo Go yes, no frame processor, about 9.9 MB packed offline or 206 kB light, on-device MoveNet in a WebView, proprietary licence, Freemium €0. react-native-mlkit-pose-detection and expo-pose-landmarks: keypoint count, Expo Go, bundle size, on-device, and price are not published; both are MIT. react-native-mediapipe-posedetection: 33 landmarks, live camera uses a Vision Camera frame processor, MIT, Expo Go and price not published. react-native-executorch: 17 COCO keypoints on YOLO26 Pose, Expo Go no, frame processor not required, MIT, model file sizes published, npm package size and price not published. Markdown: https://www.posetracker.com/compare/react-native-pose-detection.md
- [A home workout that can see you](https://www.posetracker.com/home-workout) : A home workout that can see you uses the phone camera on the person, not only the instructor video. PoseTracker is the on-device layer a fitness product puts behind that session: reps, squat form and joint angles. It does not ship a workout-video library. Not a medical device. Not the 2026 model guide. Markdown: https://www.posetracker.com/home-workout.md
- [On-Device Pose Estimation: Offline, Private, Real-Time](https://www.posetracker.com/on-device-pose-estimation) : PoseTracker runs TensorFlow MoveNet on-device. Keypoint inference does not upload video. The React Native package @pose-tracker/react-native-pose-estimation bundles the model for offline use; the light package fetches weights at runtime. An optional API key scores the pose stream (reps, angles, form) without sending the camera feed as a video file. Not a cloud GPU pose API. Markdown: https://www.posetracker.com/on-device-pose-estimation.md
- [Real-Time Pose Estimation for Fitness Apps](https://www.posetracker.com/pose-estimation-for-fitness-apps) : PoseTracker is a pose estimation SDK for fitness apps on React Native, iOS, Android and the web. On-device MoveNet keypoints are free. An API key unlocks rep counting, joint angles and a form score for catalog exercises (squats, push-ups, lunges, jump) and custom movements authored in the Assistant. Claude or Cursor can write that integration: paste https://mcp.posetracker.com/api/mcp, then Allow at https://app.posetracker.com/oauth/authorize. It is not a 3D mocap stack and not VRChat PoseTrackerVRC. Markdown: https://www.posetracker.com/pose-estimation-for-fitness-apps.md
- [Count Push-ups from the Phone Camera](https://www.posetracker.com/push-up-counter) : A push-up counter in PoseTracker is the exercise engine on on-device MoveNet, not a downloadable timer. One phone camera, one person, 17 COCO points. An API key returns valid push-up reps and a form score for a React Native, iOS, Android or web app. Plank is in the same catalog as a hold. Sit-ups are not a separate built-in counter; a 2D-visible sit-up can be described in the Assistant after sign-in. Raw MoveNet, MediaPipe and YOLO pose return keypoints only. PoseTracker is not a consumer push-up app and not a medical form check. Markdown: https://www.posetracker.com/push-up-counter.md
- [Range of Motion from the Phone Camera](https://www.posetracker.com/range-of-motion) : Computerized range of motion here means a joint angle a phone camera can see, in 2D, from PoseTracker’s on-device MoveNet (17 COCO points). It is for a fitness, yoga or rehab app that needs to show a hold: front split, bridge, and similar poses in the flexibility tool, or a side-view trunk fold in the back-flexibility test. It is not a hardware digital goniometer, not a tape Modified Schober test in centimetres, and not a table of normal clinical ranges. Hidden joints and true 3D shoulder norms are out of scope. PoseTracker does not diagnose. Markdown: https://www.posetracker.com/range-of-motion.md
- [Correct squat form from the phone camera](https://www.posetracker.com/squat-form) : PoseTracker is how a fitness or rehab product watches squat form from a single phone camera. People ask how to know if a squat is correct without a trainer. Cue lists cannot answer that. The camera can: depth, knee path, and whether the rep actually happened. This page is the movement-education job, not a model-zoo article. The phrase people use is squat form. Variants people search: proper squat form, correct squat form, goblet squat form, front squat form. Engine is on-device 2D COCO-17 MoveNet. Not medical advice. Not VRChat. Markdown: https://www.posetracker.com/squat-form.md
- [PoseTracker Assistant: Pose Estimation Integration from a Description](https://www.posetracker.com/studio) : PoseTracker Assistant is a purpose-built LLM for pose-estimation and motion-tracking integration in fitness, wellness and rehab apps. Guest mode on www.posetracker.com matches catalog requests to a copy-paste snippet. Signed-in mode at app.posetracker.com authors a V4 tracking spec (joints, metrics, phases) and a skeleton preview from a natural-language description. Claude and Cursor reach the same account through the MCP connector: paste https://mcp.posetracker.com/api/mcp, then Allow at https://app.posetracker.com/oauth/authorize. It is not a general chatbot. The guest Assistant is public on this site, without a sales call. Engine remains on-device TensorFlow MoveNet, 17 COCO. Not VRChat, not 3D. Markdown: https://www.posetracker.com/studio.md
- [Virtual physical therapy that can see you](https://www.posetracker.com/virtual-physical-therapy) : PoseTracker is the eyes inside a virtual physical therapy or home-rehab product. A physiotherapist who cannot see the patient cannot pick the next exercise or see compensation. Zoom plus a PDF is that. This page is the rehab job: virtual physical therapy, physical therapy at home, telehealth PT, PT apps. PoseTracker is not a medical device and does not replace a clinic. It gives the product movement from one phone camera (range, whether the exercise happened) so a protocol can adapt. On-device 2D COCO-17. Not the 2026 model-zoo article. Markdown: https://www.posetracker.com/virtual-physical-therapy.md

## Documentation
- [Docs hub](https://docs.posetracker.com) : official PoseTracker documentation — WebView and SDK.
- [WebView vs SDK](https://docs.posetracker.com/choose-integration) : pick the hosted tracking URL or an npm package.
- [SDK docs](https://docs.posetracker.com/quickstart) : React Native (Expo), iOS, Android and web SDK — free on-device MoveNet keypoints, optional API-key exercise engine.
- [WebView / iframe docs](https://docs.posetracker.com/webview/quickstart) : embed the camera in an existing WebView or iframe.
- [MCP connector](https://docs.posetracker.com/ai/mcp-connector) : how to use it — Claude Settings → Connectors → Add custom connector, paste https://mcp.posetracker.com/api/mcp, Connect, then Allow at https://app.posetracker.com/oauth/authorize. Cursor: the same URL in ~/.cursor/mcp.json. Claude Code: `claude mcp add --transport http posetracker https://mcp.posetracker.com/api/mcp` then /mcp → Authenticate. Article: https://www.posetracker.com/news/connect-claude-to-posetracker-mcp.

## Pricing
- [Freemium](https://www.posetracker.com/#pricing) : €0 forever free, up to 200 API calls per month, non-commercial use only.
- [Developer](https://www.posetracker.com/#pricing) : €50 per month, up to 1 000 API calls, commercial use.
- [Business Plan](https://www.posetracker.com/#pricing) : from €150 per month, no hard cap, custom quota — [book a call](https://calendly.com/fabrice-sepret/shaping-the-future-of-posetracker) to scope it.

## Blog
- [Blog](https://www.posetracker.com/blog) : Explore news, articles, and tips on various topics: computer vision, motion tracking, fitness, gaming, health, iOS and React Native development. Learn how to integrate PoseTracker into a any app for fast and accurate body movement tracking. Stay updated with our latest releases and discover how to monitor and analyze body movements easily.
- [Connect Claude to PoseTracker](https://www.posetracker.com/news/connect-claude-to-posetracker-mcp) : Connect Claude or Cursor to your PoseTracker account with the MCP connector. You click Allow; the agent then writes the tracking integration.
- [PoseTracker v2: what changed in 2026](https://www.posetracker.com/news/posetracker-v2-2026) : PoseTracker v2: on-device MoveNet SDK for React Native, iOS, Android and the web. Free keypoints. Optional API key for reps, angles and a form score.
- [Add pose estimation to a JavaScript or React app](https://www.posetracker.com/news/javascript-react-pose-estimation-sdk) : Add pose estimation to a browser app in JavaScript or React: PoseTracker web SDK, TensorFlow.js and MoveNet, free keypoints, camera, video and images.
- [PoseTracker SDK for React Native and the web](https://www.posetracker.com/news/posetracker-sdk-react-native-web) : PoseTracker SDK for React Native, iOS, Android and the web: free on-device MoveNet keypoints, an optional API key for reps, and the classic WebView API.
- [React Native pose estimation on iOS and Android](https://www.posetracker.com/news/react-native-pose-estimation-expo) : Add pose estimation to React Native and Expo: install offline or light, read free on-device keypoints, unlock exercises with an optional API key.
- [Best Pose Estimation Model in 2026: The Real-Time Mobile Guide](https://www.posetracker.com/news/best-pose-estimation-model-in-2026-the-real-time-mobile-guide) : Compare MoveNet, MediaPipe, ML Kit, YOLO-pose and Apple Vision on real-time mobile FPS. A 2026 guide to picking a pose model for a phone app.
- [Add Real-Time Pose Estimation to Mobile Apps in 2025](https://www.posetracker.com/news/how-to-add-real-time-pose-estimation-in-2025) : Integrate real-time pose estimation into iOS, Android and web apps with PoseTracker. No SDK required - motion analysis and form feedback out of the box.
- [Real-Time Exercise Analysis: Algorithmic vs ML Motion Tracking](https://www.posetracker.com/news/real-time-pose-estimation-exercise-analysis) : PoseTracker's AI motion analysis delivers real-time exercise recognition, precise pose estimation, and instant form correction for fitness app developers.
- [The Future of Fitness: AI-Powered Pose Detection and LLM Trainers](https://www.posetracker.com/news/the-future-of-fitness) : AI pose estimation and LLM trainers are transforming fitness apps with real-time feedback, personalized coaching, and seamless developer integration.
- [Flexibility Training with AI: PoseTracker Photo Analysis](https://www.posetracker.com/news/flexibility-training-with-posetracker) : AI-powered photo analysis from PoseTracker enables real-time pose detection, angle measurement, and personalized feedback for flexibility training.
- [How Real-Time Feedback Drives Client Progress and Retention](https://www.posetracker.com/news/the-role-of-real-time-feedback-in-maximizing-client-progress-and-retention) : Boost client progress and retention with AI-powered real-time feedback. PoseNet and MoveNet deliver scalable form correction for fitness apps.
- [How AI and Motion Tracking Will Transform Personal Training](https://www.posetracker.com/news/how-ai-and-motion-tracking-will-transform-personal-training) : AI and motion tracking are revolutionizing personal training with real-time feedback, personalized workouts, and tools like MoveNet and PoseNet.
- [Personalized Feedback to Boost Client Retention in Fitness Apps](https://www.posetracker.com/news/how-personalized-feedback-enhances-client-retention-in-fitness-apps) : Personalized feedback in fitness apps drives client retention through motion tracking, data-driven insights, and tailored workout plans powered by AI.
- [AI-Powered Fitness Coaching: Revolutionizing Online Training](https://www.posetracker.com/news/revolutionizing-online-fitness-coaching-with-ai-powered-fitness-coach) : PoseTracker revolutionizes AI fitness coaching with real-time pose estimation. Build hyper-personalized training experiences across fitness platforms.
- [Best Human Pose Estimation Models for Mobile App in 2024](https://www.posetracker.com/news/best-human-pose-estimation-models-for-mobile-app-in-2024) : Compare MoveNet, PoseNet, BlazePose, YOLO, and MLKit for mobile apps. Discover how PoseTracker API adds real-time pose estimation to any application.
- [Enhance Your App with Real-Time Pose Estimation](https://www.posetracker.com/news/real-time-pose-estimation-integration) : PoseTracker delivers real-time pose estimation via MoveNet for high accuracy. Ideal for developers building fitness, gaming, and interactive apps.

## Blog categories
- [AI Yoga Poses](https://www.posetracker.com/article-categories/ai-yoga-poses) : PoseTracker blog articles filed under this category.
- [Apps](https://www.posetracker.com/article-categories/apps) : PoseTracker blog articles filed under this category.
- [Development](https://www.posetracker.com/article-categories/development) : PoseTracker blog articles filed under this category.
- [Fitness](https://www.posetracker.com/article-categories/fitness) : PoseTracker blog articles filed under this category.
- [Future Of Fitness](https://www.posetracker.com/article-categories/future-of-fitness) : PoseTracker blog articles filed under this category.
- [LLM Fitness Trainer](https://www.posetracker.com/article-categories/llm-fitness-trainer) : PoseTracker blog articles filed under this category.
- [Pose Estimation](https://www.posetracker.com/article-categories/pose-estimation) : PoseTracker blog articles filed under this category.
- [Tech](https://www.posetracker.com/article-categories/tech) : PoseTracker blog articles filed under this category.

## Optional
- [Privacy Policy](https://www.posetracker.com/privacy-policy) : PoseTracker Privacy Policy
- [PoseTracker API: Terms of Use](https://www.posetracker.com/terms-of-use) : PoseTracker API Terms of Use
- [Legal notice](https://www.posetracker.com/legal-notice) : publisher Movelytics SAS, Bidart, France. Email contact@posetracker.com. Phone +33 6 45 20 01 91.