FruizNet Using an Efficient Convolutional Neural Network

Fruits classification which is a crucial aspect of the agriculture industry poses a challenge of time-consuming task of manual fruits classification by farmers and industry peoples, therefore an automation is required to perform the job. This paper introduces the deep learning model to identify and categorize the freshness and rottenness of fruits. The dataset used for the classification task consists of eight types of fruits (i.e., Apple, Guava, Banana, jujube, Strawberry, Grape, Orange, and Pomegranate) each having two categories (i.e., Fresh and Rotten) are fetched from the Mendeley repository. The convolutional layers of the model extract the essential patterns from fresh and rotten fruit image samples .and to classify them into respective classes of fresh and rotten, Softmax is used as a classification function because it has multiclass and predicted class based on the class with the highest probability. The performance of the proposed model attained an overall accuracy of 99.94%. The result shows that our model can easily identify and classify fresh and rotten fruits. The proposed model was also compared with some pre-trained models like VGG16, MobileNet, XceptionNet, and EfficientNet to suggest the best and most efficient model for the task of fruit classification.