Senswork Computer Vision System
Industrial computer vision: Baumer smart cameras, Jetson boards, OpenCV C++ pipelines, controlled lighting, and reflective transformation imaging.
Problem
Industrial inspection requires consistent imaging across reflective and complex surfaces — hard to do with a single light source and a generic camera.
My Role
Computer vision engineer (internship), Senswork — Germany.
What I Built
Embedded C++ pipelines running on Jetson, integrated with Baumer industrial cameras and a 64-light LED dome to capture PTM/RTI imagery for surface defect analysis.
Key Features
- Baumer camera integration
- Jetson / ARM64 deployment
- OpenCV C++ pipelines
- Cross-compilation for embedded Linux
- 64-light LED dome imaging
- PTM / RTI generation
- Normal map / surface analysis
- CUDA edge detection
Technical Details
Cross-compiled for ARM64 Linux. Used OpenCV + CUDA for edge detection. Multi-light capture sequenced over Baumer neoAPI. Normal maps generated from the light stack.
Hardest Challenge
Cross-compilation + embedded debugging on Jetson while keeping the LED-dome capture sequence in lockstep with the camera shutter.
Outcome / Result
Production-grade pipeline that produced consistent surface analysis across reflective parts.
Learnings
Embedded vision lives or dies on lighting and timing. Algorithms come third.
Pipeline
Camera → Jetson → LED dome sequence → image processing → output normal/surface maps.
- Baumer neoAPI ingest
- 64-light LED dome stack
- OpenCV + CUDA processing on Jetson
- PTM / RTI generation pipeline