Convolutional Neural Network Based Image Processing Model for Supply Chain Management

India follows China as the world's top fruit producer. A variety of factors faced at various stages along the supply chain cause 30–33% of the produced produce to be wasted annually. One of these factors is the storage and transportation of substandard fruit. In order to classify and grade fruits, this work intends to develop an efficient and highly accurate image processing model. To do this, we created a base Convolutional Neural Network (CNN) model and compared it with the modified pre-trained ResNet models. On the basis of the selected performance criterion, a thorough study of the models and comparison of them was conducted. All of the pre-trained ResNet models outperformed our base 3-layer CNN model, which had 83.8% accuracy, with Resnet18 and Resnet34 achieving the maximum accuracy of 97.30%. The created model can be incorporated into a real-time image processing system to guarantee that quality standards are adhered to across the whole supply chain.