VivaKitAI Yoga & Exercise Posture Correction Assistant
Real-time computer vision and kinematic biomechanics coaching engine powered by Google MediaPipe Pose Landmarker (Apache 2.0). Automatically detects 33 3D full-body landmarks, recognizes 6 canonical yoga asanas (Tadasana, Vrikshasana, Warrior II, Chair, Downward Dog, Triangle), computes scale-invariant 3D joint angles, and provides instantaneous color-coded feedback and actionable posture correction directives. Runs 100% on CPU with zero GPU dependencies and zero viral copyleft licensing.
Secure checkout by Dodo Payments (UPI, cards). Your personalised download link is e-mailed within 24 hours. One purchase covers one project team of up to 4 students. See terms and refunds.
Our honest results
| Asana Recognition Accuracy | 6/6 on a small real-photo test set (one Wikimedia photo per asana) plus a 30-configuration synthetic robustness benchmark |
| Mean CPU Inference Latency | 60.8 ms (~16.5 FPS real-time throughput on standard CPU) |
| Scoring Sensitivity Correlation | -0.881 (synthetic perturbation benchmark 0° to 40°) |
| Classification Stability | 80.0% (synthetic perturbation benchmark) |
Evaluated on real held-out photographic instances from Wikimedia Commons (CC BY 3.0 / CC BY-SA 4.0) and 30 synthetic angular perturbation test configurations. Includes Gradio 6 console with interactive joint angle table, automated pytest suite, and 15 viva voce questions on 3D spatial geometry, BlazePose architecture, and biomechanical tolerances.
What's included
- Readable source codeSmall, commented modules you can explain line by line in your viva.
- Demo appA local web app to show your examiners - runs offline on your laptop.
- Trained model + reference resultsSo you can compare your own runs fairly.
- Learning guide with viva Q&ABackground, design choices explained, limitations and extension ideas.
- One-click setup & testsWindows setup script and tests that prove it works on your PC.
- Dataset download scriptFetches the public data from its original source, with the citation.
Dataset credit
Google MediaPipe Pose Landmarker (Apache-2.0) and reference posture imagery from Wikimedia Commons (CC BY 3.0 / CC BY-SA 4.0 by Nina Mel and Zamkovoi).
Academic integrity
We sell code and teaching, not finished projects. You must understand, extend and write up the work yourself, acknowledge the kit and cite the dataset. No ready-made reports.