PocketPose

Motion-Driven Experiences Now in Your Pocket

Getting Started

Develop interactive applications, sports analysis tools, or creative motion-powered experiences with our lightweight library that integrates seamlessly into your project.

Get Started Now

PocketPose takes care of real-time motion estimation, so you can focus on building what matters.
Get started with our Python API today!
JavaScript and Kotlin APIs coming soon

Effortless Integration

No need to be an expert in pose estimation. Just plug in PocketPose and start using body, hand, and face motion data instantly.

Optimized Performance

Designed for mobile and edge devices, ensuring smooth real-time inference without heavy computational requirements.

Built for Developers

Simple APIs, clear documentation, and an active community to help you bring motion-powered features to your users faster.

For Researchers

Knowledge Hub: Learn Human Pose Estimation

Whether you're new to Human Pose Estimation (HPE) or an experienced researcher, our Knowledge Hub is designed to help you explore and deepen your understanding of HPE concepts, methods, and applications.

We provide structured learning resources at all levels, from foundational concepts to advanced techniques. Start with the Basic Concepts to grasp core principles, then explore topics like deep learning-based pose estimation, 3D human pose recovery, multi-person tracking, and real-time implementation.

Explore the Knowledge Hub

If you find the PocketPose library or the Knowledge Hub useful in your research, please consider citing our work:

    
    @misc{khan2025pocketpose,
          title={PocketPose: Real-Time Human Pose Estimation on Consumer Devices - A Comprehensive Study}, 
          author={Muhammad Saif Ullah Khan},
          year={2025},
          journal={arXiv preprint arXiv:2503.16345},
          eprint={2503.16345},
          archivePrefix={arXiv},
          primaryClass={cs.CV}
    }