Federated Learning for Mango Ripeness Classification: Evaluating Pre-Trained Deep Learning Frameworks

Classifying mangoes for maturity is crucial work in contemporary agriculture since it affects harvesting and distribution methods. To solve the problem of classifying the maturity of mangoes, we use federated learning, a distributed machine learning approach. Using a ripening mango data set, we compare the performance of several pretrained deep learning frameworks, including VGG11, AlexNet, SqueezeNet, ShuffleNet, ResNet18, and InceptionNet. The experiment simulates a federated learning environment by using a wide variety of client sizes, from 3 to 7. Findings reveal InceptionNet to be the top performer, with a remarkable 98.75% accuracy rate in categorization. The results of this research show that federated learning may improve the accuracy with which mango ripeness is classified, providing valuable information for the development of agricultural methods.