Classification of Agricultural Fruit Imagery using CHAR

The use of Deep Learning for Computer Vision in the fields of fruit and agricultural image classification is an intriguing and emerging field of research. This approach reduces time and replaces manual work of fruit classification. The implementation proposed in this paper, CHAR, involves the use of a Convolutional Neural Network (CNN) for fruit classification trained on an image dataset following hyperparameter optimization, specifically Bayesian optimization. Autoencoders are utilized in this model for denoising purposes. Subsequently, regularization is done in order to ensure that overfitting or underfitting does not occur. The results obtained displayed a classification accuracy of 98.96%.