"Optical Lego Recognition" is a computer vision and machine learning category (part of CBIR or "Content-based image retrieval") whereby a computer can recognise lego pieces from still or moving images of lego blocks. (Moving, because they may be on a conveyor belt, or I guess they may be being thrown at the camera, by a lego-wielding adversary)

Picture this:

You use an app to take a photo of a pile of Lego pieces.

The app recognises every single piece of lego (and some that are not lego) and from that set suggests some cool things to make.

A more pedestrian use of OLR is for sorting lego into bins, making it easier to sell.

External References

• Botha et al.: Realtime LEGO Brick Image Retrieval with Cellular Automata (PDF)
• Cole et al.: Visual Object Recognition using Template Matching (PDF)
• jacquesm: Sorting 2 Metric Tons of Lego
• HN: Sort Lego