Karevo is designed for agricultural operations that spend significant time on manual quality sorting of potatoes. Particularly for direct marketers and family farms who want to make their sorting more efficient and precise.
AI-Powered Potato Sorting
Retrofittable
Compatible with all standard reading tables on the market. Existing infrastructure can continue to be used. Fast implementation and ready for immediate use.
Image recognition with
artificial intelligence
Reliable defect detection. Defect severity can be individually adjusted.

Fully automatic sorting
Throughput of up to 5 t/h. Statistics on potato quality, can be monitored online.
Detect and sort various potato defects

Growth crack

wireworm

drot

malformed

Mechanical Damage

Wet Drot

Greening

Developed for family farms and direct marketers.
History

10/2021
First Idea

03/2022
Concept Development

10/2023
Exist Gründerstipendium

05/2024
Prototype Development

03/2025
Field Test

10/2025
Go to Market
Events
Am Staudengarten 2, 85354 Freising
Sales Partner
General
Karevo was developed for quality sorting after storage of table potatoes, directly before sale to end customers.
No oversized system, but a solution with clear strengths: Robust. Simple. Flexible. Fair pricing.
Installation
Depending on the variant and individual requirements, installation is typically completed within one day. There may be slight variations depending on the operation and requirements.
Currently, the system has two outputs - one for rejected goods and one for saleable goods. Models with more outputs are in development.
No, no connection is needed for regular operation. However, internet access is required for remote maintenance and updates (e.g., via WiFi or hotspot).
Yes - provided the system is connected to the internet.
Performance
Currently, unwashed, yellow-skinned potato varieties can be sorted. Red potatoes will be available in the future..
Currently, the focus is on potatoes. Expansion to products like carrots or onions is planned.
The Karevo system can sort up to 5 tons per hour. The throughput depends on the uniformity of the potatoes. If they are misshapen or very large, the belt must be slowed down so that detection still functions optimally.
The following defects can be detected: green discoloration, wet rot, dry rot, wireworm damage, growth cracks, mechanical damage. Clods, stones, and deformations can also be sorted out.
The detection accuracy for defects is approximately 90%.
Yes. Karevo’s AI models have been adapted and trained for unwashed goods. Like the human eye, detection improves the cleaner the tubers are.
We support you during setup so that Karevo is optimally adjusted to your requirements. Through targeted calibration, sorting accuracy can be continuously improved. A completely error-free sorting is not technically possible, as with any automated system.
Operation
Yes. If, for example, higher throughput rates are desired or a variety is not yet recognized, the optical system can be deactivated and Karevo can be operated like a normal conveyor table.
Yes. Both the sorting sensitivity and the defect size are individually adjustable.
Our team is on-site for commissioning for approximately 3 days. The system is individually adjusted to your operation and you receive comprehensive training.
No. The user interface is intuitive and user-friendly - a brief introduction is sufficient.
Miscellaneous
AI is a technology that enables machines to learn from data and make independent decisions - similar to the human brain. With Karevo, AI recognizes defects quickly and precisely and continuously improves.
Yes. Karevo uses machine learning and continuously improves through new training data and software updates.
We are in contact with various funding agencies. We will provide information on this shortly.
Partner




