Industry 4.0 Meets Gastronomy: Enhancing Pandang Cuisine Classification with CNNs and Transfer Learning

The study article investigates the utilisation of Deep Convolutional Neural Networks (CNNs) and Transfer Learning in the domain of Pandang Cuisine Classification. The objective is to improve effectiveness and profitability in the food and beverage business by using AI technology. The research acknowledges the importance of Industry 4.0 and used a dataset consisting of 993 photographs categorised into 9 kinds of Pandang meals. By using the MobileNetV2 pretrained CNN model, the approach entails loading, transforming, and preprocessing the data, and then training the model using callbacks such Model Checkpoint and Early Stopping. The hyperparameters consist of a batch size of 32, 100 epochs, and an output layer including 9 classes. The model attains an impressive accuracy of 90% on the test dataset, taking into account supplementary assessment criteria like as precision, recall, and F1 score. The study culminates in the depiction of accuracy and loss curves, forecasts on the test data, and the exhibition of Classification Reports and Confusion Matrix, complemented with Grad-Cam visualisations. This study shows potential for enhancing culinary processes and optimising supply chain management in the food business.