During my university studies I worked at Adaptive Vision, a small Polish company building software for camera-based industrial quality inspection.

I worked there for three years (2009–2012) and ended up becoming the engineering lead for the image recognition team. It was a fantastic adventure: we were working with real end-to-end systems, building solutions for real problems of our clients in factories around the world.

At the end of my time at Adaptive Vision I wrote a small book summarizing what I learned about the various image analysis techniques, based on 3 years of experience applying them to realistic industrial quality inspection problems.

The book, which served a dual purpose as my CS master thesis, is called “Image Analysis Techniques for Industrial Inspection Systems”. It covers:

  1. Image Thresholding
  2. Blob Analysis
  3. 1D Edge Detection
  4. 2D Edge Detection
  5. Contour Analysis
  6. Shape Fitting
  7. Template Matching

You can find it here . The company used to send a printed copy to all customers along with their license.

Addendum (2025)

Some notable things that happened since then:

  • the company got acquired and rebranded as Zebra Aurora Vision . The product is still out there and they seem to be doing well!
  • the Alexnet paper came out the year I left Adaptive Vision. I got to experience the last moments of “good old-fashioned AI” in image recognition, just before deep learning solutions became state-of-the-art.