Identifying Mango and Its Ripeness Using Image Processing and Machine Learning Approach

Abstract

Fruit market is a subject of choice, thereby, a dealer needs to grade the fruit.  Fruit grading commercially available systems are very expensive, and manual fruit grading systems used in small businesses and dealers are prone to human error and inaccuracy. This paper proposes a system for identifying and grading Mango which will bebeneficial if we consider Industry 4.0. A Faster Region-based Convolutional Neural Network (Faster R-CNN) object detection algorithm using Tensor Flow has been implemented for identifying the fruit and by Image processing the probable percentage of ripeness can be determined. Thereby categorizing the fruit into classes. The results show that the proposed methods are efficient and cost-effective for determining and detecting the ripeness of fruits. The same system, when trained effectively can be used for multiple fruits.