In the realm of high-tech greenhouse cultivation, the integration of artificial intelligence (AI) into the sorting process of pears presents a noteworthy advancement. Cees Maessen, a former data science student, delved into this topic for his master's thesis, aiming to enhance the efficiency and consistency of pear sorting. This process is crucial, as it determines which pears are suitable for supermarket shelves and which are not, due to defects such as dents, black streaks, or diseases.
The sorting of pears, traditionally reliant on manual labor and prone to human error, is conducted by specialized companies using conveyor belts and water channels. These companies face the challenge of excluding unsuitable pears while minimizing food waste. Maessen's research focused on employing computer vision powered by AI to refine this process, reduce manual input, and improve sorting accuracy.
Maessen initiated his research by establishing an infrastructure for data collection and processing. Utilizing advanced cameras, the study captured images of pears from various angles. The AI model employed in this study was trained to autonomously recognize patterns and characteristics of pears, such as shape, color, and texture, without the need for all images to be manually labeled. This model was further trained with labeled data to distinguish between categories like 'rotten' or 'not rotten'.
The findings from Maessen's research indicate that AI has the potential to significantly enhance the pear sorting process. The developed model can more accurately assess the suitability of pears for different uses, and it has shown capability in detecting issues such as rotten spots, irregular shapes, and colors indicative of diseases or damage. This advancement in AI technology not only improves the reliability and consistency of the sorting process but also holds promise for precision agriculture by providing valuable feedback to growers.
This research addresses critical societal and economic challenges, including labor shortages in manual fruit sorting and the ongoing battle against food waste. By improving sorting accuracy through AI, the study contributes to preserving more fruit and potentially applying similar techniques to other crops. Looking forward, Maessen's work lays the groundwork for further innovations, including the possibility of achieving full traceability from tree to pear, thereby enhancing transparency and sustainability in agriculture.
Source: Univers