DFNet

March 17, 2025 | Sustainability |

To develop an automated classification system for dragon fruit varieties using machine learning techniques, researchers from Akdeniz University, Türkiye, and the National University of Science and Technology Politehnica Bucharest, Romania, conducted a study. Accurate classification of agricultural products like dragon fruit is essential for quality control, efficient logistics, consumer satisfaction, and sustainability. With rising global demand, automation in sorting and packaging has gained importance to reduce manual labor and enhance operational efficiency.

The study focused on classifying four commonly cultivated dragon fruit varieties—American Beauty, Dark Star, Vietnamese White, and Pepino Dulce—based on measurable color, mechanical, and physical attributes. Data were collected from 224 fruit samples using digital image processing, colorimetry, electronic weighing, and stress–strain testing to ensure objective and reproducible measurements.

Three machine learning models—Random Forest, Gradient Boosting, and Support Vector Classification—were tested for their classification performance. Among them, the Random Forest model achieved the highest accuracy at 98.66%, showing strong performance across all evaluation metrics. This was attributed to its capability in handling nonlinear data patterns and reducing overfitting through ensemble learning.

The study demonstrates the potential of machine learning in fruit classification, while also noting the need to address challenges such as environmental variability and genetic differences in future research.

Read more