Fruit recognition using image processing

Abstract

Manually classifying and evaluating anything is difficult. It is difficult to manually count ripe fruits and evaluate their quality. Increasing labor costs, a shortage of skilled workers, and declining storage costs are just some of the major challenges associated with fruit production, marketing, and storage, among others. An effective method for localizing all clearly visible objects or portion of an object from an image has been proposed in this study, requiring less memory and processing resources. The main obstacles for object detection, such as object overlap, background noise, low resolution, etc, that prevents us from obtaining better results has been overcome by processing every input image. It also built an enhanced classification or recognition algorithm based on convolutional neural networks, which has shown to perform better than baseline studies.