Strawberry Image Segmentation Based on U^ 2-Net and Maturity Calculation

Abstract

Strawberries are one of the most important cash crops and are widely grown around the world. Since strawberries have a short ripening period and not all strawberries are of the same maturity, it is important to know the specific maturity value of each strawberry in a timely and accurate manner for automatic strawberry picking. This study is aimed at image processing in order to achieve numerical values of strawberry maturity as a quantitative indicator of strawberry maturity. A deep network U^2-Net with significant object detection is used, trained and tested to automatically segment strawberries and background in the image; Two-Pass concatenated domain analysis is used to segment individual strawberries in the mask, and then the percentage of red pixels in the segmented individual strawberries is calculated.