Smart farming with a systematic approach: Cost-effective automation using image processing for family-run businesses

Challenge: Ensuring quality, reducing costs

Sorting potatoes and onions by hand is time-consuming and labour-intensive. Various types of damage, such as growth cracks, wireworm infestation, dry and wet rot, deformities, mechanical damage or green discolouration, can only be reliably identified with considerable effort.

Benedikt Keßler, co-founder and managing director of Karevo, emphasises: "The Karevo DUO85 was developed with the aim of making optical sorting technology accessible to direct sellers and family-run businesses. This enables even smaller businesses to embrace digitalisation and automate their processes efficiently. The system ensures consistently high product quality and reduces the workload for staff during sorting."

Solution: AI-based vision technology for potatoes and onions

The Karevo DUO85 fully automates the sorting process. Depending on the system configuration, between two and eight GigE Vision cameras from the IDS uEye FA series are used. These capture every tuber from every angle and provide a complete 360° view of its surface. The system takes over 15 images per potato and analyses them in real time using a high-performance computer.

The image data is transmitted to a specially developed software program, which uses artificial intelligence to assess characteristics such as size, defect type and severity, shape and other quality parameters. The AI detects the seven most common quality defects with pixel-level precision and up to 95% accuracy, whilst simultaneously assessing their intensity and size. In addition to potatoes, the DUO85 also sorts yellow and red onions, including reliable foreign object detection: Stones and clods are reliably removed from both potatoes and onions.

The AI models used are designed for stable operation in real-world conditions and can be further optimised using additional image data where necessary, for example, in the case of heavily soiled crops or specific varieties.

Based on this analysis, the system controls a pneumatic finger system – a compressed-air-powered diverting and sorting system in which individually controllable, finger-like actuators divert products from the conveyor stream in a targeted and gentle manner, synchronised with the image processing. This ensures that damaged tubers or those that do not meet specifications are reliably separated.

The result: Consistent sorting results, reduced staffing requirements and a stable, continuous throughput of up to 5 tonnes per hour – a clear economic advantage for businesses of all sizes.

Economic benefits: Getting started with a manageable investment

A key objective in the development of the DUO85 was to ensure a low barrier to entry. The system is specifically designed for direct sellers and family-run businesses, which have previously had little access to AI-powered image processing and vision technology. "With the DUO85, we’re making it possible for smaller businesses to get started with digitalisation," says Keßler. "Our customers benefit from consistently high product quality, a significant reduction in the workload on staff and more efficient processes – without high investment costs."

Profitability always depends on individual operational structures. Karevo typically estimates that one to two (temporary) staff members can be saved, which in many cases results in a payback period of around two to three years.

The pricing is specifically designed for small and medium-sized businesses to optimise return on investment and ensure profitability. Furthermore, the system is modularly scalable and can be flexibly adapted to businesses of different sizes – an important factor in ensuring long-term investment security.

Camera technology from IDS: Robust, powerful and easy to integrate

Karevo uses IDS uEye FA series GigE Vision cameras featuring the 5.10 MP Sony IMX547 sensor from the Pregius S series. Thanks to global shutter and BSI (back-side illumination) technology, these cameras deliver excellent image quality, even in challenging lighting conditions in sorting and processing areas. The robust design, featuring IP69K protection, screw-lock connectors and an extended temperature range, makes these cameras ideal for use in harsh agricultural environments. Features such as on-camera pre-processing also reduce the computing power required.

Jürgen Hejna, Product Manager 2D Machine Vision at IDS, describes it as follows: "For applications such as the DUO85, reliability, robustness and ease of integration are key factors, alongside image quality. The uEye FA cameras offer precisely these features and enable reliable multi-camera operation whilst minimising integration and maintenance requirements."

Quick integration and future-proofing with IDS peak

Another economic benefit stems from the use of the IDS peak SDK. "Using IDS peak enables us to integrate the cameras into our system quickly and reliably. Thanks to optimised control, parameterisation and diagnostic functions, we can ensure high image quality and reliable performance in agricultural applications. At the same time, we benefit from long-term compatibility and reduced development costs for future system upgrades," explains Benedikt Keßler.

The user-friendly interface makes complex AI technology easy to use. Parameters can be adjusted intuitively, different sorting profiles can be switched easily, and detailed reports are available at the touch of a button. The DUO85 can be seamlessly integrated into existing sorting lines and operates as a standalone optical sorting system tailored to individual requirements.