CryoVision

Computer vision for freezer boxes: reads a 10x10 cryobox from a single photo using OpenCV and Claude's vision API, and exports a JSON map of every grid position (A1-J10) with whatever label text, cap colour or description it can make out — instead of 20 minutes of digging for one tube. Later folded into Benchmate as the step that reads freezer boxes during protocol design.

See the code →