TUM Startup Develops AI-Powered Potato Sorting Machine

Karevo, a spin-off from the Technical University of Munich (TUM), has developed an artificial intelligence-powered machine designed to automate potato sorting and process up to ten tons of potatoes per hour.

The system utilizes optical recognition and a trained AI model to achieve 95 percent accuracy in identifying defects and foreign materials. The model, trained on over 100,000 images, can detect seven types of potato defects, including rot, cracks, and damage from wireworms.

A key feature of the system is its ability to sort unwashed potatoes. Karevo co-founder Benedikt Keßler stated that many existing detection methods struggle with unwashed potatoes due to variations in appearance based on soil, region, and storage conditions. The AI model can be quickly adapted to specific farm conditions.

The company specifically targets small and family-run farms with its machine, citing that many existing solutions are too large, expensive, or difficult to maintain for smaller agricultural businesses. Karevo's system is designed to be smaller, more affordable, and modular for easier integration and maintenance.