Quality Assessment of Orange Fruit Images Using Convolutional Neural Networks

For sustainability of life and rural development, assessment of fruits in a non-destructive manner is required. The fruits which are available in the market must fulfill buyer needs. Orange fruit analysis is generally done by visual examination and by observing the size. For large volumes, we cannot assess by human graders, so picture preparation is done on quantitative, solid, and predictable data. This research contains a division of orange images based on fresh and rotten criteria, the images are of different kind based on the rotation of images. Then the classification of images is done based on CNN, binary cross-entropy loss function, along with accuracy is calculated and resultant graphs are illustrated with an accuracy of 78.57%.