Role overview
Designs and implements the visual feature extraction pipeline, ensuring high-quality input data for the ML model from multi-camera capture system
Responsibilities
- Design and deploy 3-camera capture system (top-down + 2 oblique)
- Implement cross-polarized lighting setup for glare elimination
- Develop visual feature extraction algorithms: Skin blanching detection Contact patch area measurement Finger flexion analysis (keypoint tracking) Micro-tremor detection (10-20 Hz)
- Skin blanching detection
- Contact patch area measurement
- Finger flexion analysis (keypoint tracking)
- Micro-tremor detection (10-20 Hz)
- Synchronize camera streams with hardware frame-lock
- Collect and curate training dataset (100+ matches)
- Optimize feature extraction for real-time performance (<8ms budget)
- Implement confidence scoring for feature quality
- Handle challenging conditions (varied lighting, athlete positioning)
- Support broadcast integration with visual debugging tools
- Refine calibration procedures based on demo feedback
- Implement failover and redundancy for camera failures
- Optimize for 98%+ uptime during live events
- Develop automated quality monitoring and alerting
- Support LED synchronization (Art-Net/DMX integration)
- Production-grade error handling and recovery
- 5+ years experience in computer vision engineering
- Expert-level knowledge of OpenCV and image processing techniques
- Experience with high-speed camera systems (120+ FPS)
- Strong understanding of optical phenomena (lighting, polarization, color science)
- Experience with multi-camera synchronization and calibration
- Proficiency in C++ and Python for real-time CV pipelines
- Experience with GPU-accelerated image processing (CUDA, cuDNN)
- Experience with industrial vision systems or broadcast/entertainment applications
- Knowledge of color-based feature extraction (blanching, perfusion analysis)
- Experience with pose estimation and hand/finger tracking (MediaPipe, OpenPose)
- Background in optics and lighting design for machine vision
- Experience with GigE Vision or USB3 Vision camera protocols
- Familiarity with embedded vision systems or edge deployment
Preferred qualifications
- Experience with NIR imaging or multi-spectral cameras
- Knowledge of photogrammetry and 3D reconstruction
- Experience with motion capture systems or sports analytics
- Background in signal processing for vibration/tremor detection
- Familiarity with broadcast equipment and professional video workflows
- Hands-on hardware expertise: Comfortable with physical camera setup and troubleshooting
- System thinking: Understand end-to-end pipeline from optics to ML model
- Attention to detail: Ensure data quality and consistency across diverse conditions
- Pragmatism: Balance theoretical perfection with practical constraints (time, budget)
- Field readiness: Willingness to travel for on-site deployments (2-3 trips to US)
Tags & focus areas
Used for matching and alerts on DevFound Machine Learning Computer Vision Ai