We are seeking a talented Computer Vision Engineer to design and develop mirror-aware, multi-camera perception systems for real-time human pose tracking. The role involves building robust 2D/3D pose estimation pipelines, handling occlusions and reflections, and optimizing performance for wearable and mobile devices.
Key Responsibilities
- Design and implement multi-camera and mirror-aware computer vision architectures.
- Develop 2D and 3D human pose estimation systems.
- Implement camera calibration, stereo vision, and multi-view reconstruction pipelines.
- Perform mirror detection and reflection analysis using geometric consistency techniques.
- Develop algorithms for 3D joint triangulation and reprojection validation.
- Build real-time biomechanical analysis systems for posture, movement, and form correction.
- Implement temporal filtering and synchronization between camera streams.
- Optimize inference pipelines to achieve sub-100ms latency.
- Support ONNX and TensorFlow Lite (TFLite) model deployment.
- Create evaluation datasets, performance metrics, and automated testing frameworks.
- MediaPipe
- YOLO
- HRNet
- ONNX
- TensorFlow Lite (TFLite)
- Edge AI
- On-Device Inference
Preferred Qualifications
- Experience with multi-camera or stereo camera systems.
- Production deployment of Computer Vision applications.
- Experience with Edge AI and mobile deployment.
- Exposure to AR/VR, Robotics, Sports Analytics, or Fitness Technology.
- Familiarity with CVPR, ICCV, or ECCV research literature.