Comparative Analysis of Convolution Neural Network Models for Automated Grading of Muskmelons

Muskmelons are widely appreciated for their taste and nutritional benefits, but manual grading of muskmelons based on visual quality is time consuming, subjective, and prone to inconsistency. To address this challenge, the present study explores the application of deep learning for automated fruit grading. Three pre-trained convolutional neural network models, namely DenseNet201, InceptionV3, and MobileNet, were evaluated using a dataset of 616 segmented muskmelon images categorized into two quality classes. The models were finetuned and assessed based on classification accuracy, precision, recall, and F1-score. Among the evaluated models, MobileNet achieved the highest performance with an accuracy of 87 percent, offering an optimal balance of accuracy and computational efficiency. Its lightweight architecture makes it suitable for realtime deployment on resource constrained agricultural devices. These findings demonstrate the potential of convolutional neural networks to enhance consistency and efficiency in fruit quality assessment within the agricultural supply chain.