Font size:

Intelligent Detection Technology for Pineapple Quality

Author:Taiwan Agricultural Research Institute,COA

In this work we use hyperspectral images combined with AI technology to develop non-destructive detection methods to establish three quality identification models for internal browning of pineapple tissues, bruise, and sound of fruit, which assists the industry in grading the quality of export. The internal browning detection technology of pineapple cold damage is based on the summer fruit to establish a detection model, and its training accuracy rate is 89.4%. After using the winter fruit to cross-validation, the verification accuracy rate reaches 80%. Applying the bruise detection technology, the fruit can be revealed and eliminated from the hyperspectral image on the first day of injury. In the detection mode of hollow sound and solid sound fruits, the original prediction result shows that the detection rate of solid sound fruit is 79.2%, and the overall accuracy reaches 78%. After adjusting the weight coefficient, the capture rate of solid sound fruit can be achieving 93% accuracy, avoiding the impact of solid sound fruit on shipping quality.