Semantic Segmentation of Packaged and Unpackaged Fresh-Cut Apples Using Deep Learning

Computer vision systems are often used in industrial quality control to offer fast, objective, non-destructive, and contactless evaluation of fruit. The senescence fresh-cut apples is strongly related the browning pulp rather than properties peel. This work addresses identification selection inside images apples, both packaged unpackaged; this a critical step towards computer system that able evaluate their internal properties. A DeepLabV3+-based convolutional neural network model (CNN) has been developed for semantic segmentation task. It proved be robust with respect similarity colours between peel pulp. Its ability separate from background verified on four varieties apples: Granny Smith (greenish peel), Golden (yellowish Fuji, Pink Lady (reddish peel). achieved an accuracy greater 99% all these varieties. approach was isolate regions significantly affected by process unpackaged pieces: areas, colour analysis will studied products.