UP Mindanao researchers develop AI checking tool for durians

Researchers from the University of the Philippines (UP) Mindanao have developed an artificial intelligence (AI) tool to help identify and grade which durians have the best quality.

According to the Department of Science and Technology (DOST), the Philippines produces at least 80,000 metric tons of fresh durian, which is worth about P6 billion. Rating the quality of the fruit, however, has usually relied on traditional methods such as visual observation and tapping the fruit.

"These methods are subjective and difficult to replicate consistently; grading can vary among classifiers and may become a bottleneck when skilled personnel are limited, particularly during peak handling periods," project leader Dr. Armacheska Satina said in a statement.

Satina and her team from UP Mindanao developed a modernized grading system called the AI-based Non-Invasive Grading System for Durian (AIDurian).

The AIDurian makes use of computer-operated visual scanning, acoustic analysis, and deep learning to analyze the fruit's ripeness, surface defects, locule count, and shape.

"It provides a technology-enabled complement or alternative to traditional manual tapping and visual inspection, helping improve the speed and consistency of grading durian," Satina said.

AIDurian integrates two prototype platforms, the Inspection of Durian Using AI (INDAI) and the Durian Desktop-Oriented Non-Invasive Grading System (DuDONG). INDAI is used for field-level inspection and DuDONG is a desktop application for more comprehensive analysis.

The technology is said to have an 80% accuracy rate for its maturity classification and acoustic ripeness, 93% for defect detention, and 98% for percent locule analysis accuracy rates.

"By automating selected grading tasks, AIDurian can help support more timely sorting, handling, and marketing decisions while potentially reducing postharvest losses," Satina said.

"AIDurian supports the digital transformation of the durian industry, strengthens local capacity in AI applications for agriculture, and provides a potential pathway toward more standardized and sustainable grading practices," she added. — VDV, GMA News