MASK REGION CONVOLUTION NEURAL NETWORK WITH DISCRETE WAVELET TRANSFORM FOR INTELLIGENT DETECTION AND CLASSIFICATION OF FRUITS FRESHNESS

Abstract

Fruits are considered a significant agricultural crop used in numerous fields. Freshness is a key fruit quality characteristic that has a direct impact on consumers' physical health and purchasing intentions. Additionally, it plays a vital role in determining market pricing. Therefore, rapid automatic classification was used to ensure the quality of the fruits. The current work uses a deep learning-based algorithm to classify fruit freshness. This study demonstrates how to classify fruits into (Apples, Bananas, and Oranges) and determine whether they are fresh or rotten using Discrete Wavelet Transform (DWT) and MASK Region Convolution Neural Network (RCNN) with instance segmentation. First, gather images of fresh and rotten fruits (Apple, Banana, and Orange) then apply image enhancement and Discrete Wavelet Transform on the collected images then label them using make sense label. The experiment results after training show that the suggested algorithm in the classification of fruits is effective with a 99% accuracy rate.