Application of Convolutional Neural Networks in Logistics Engineering and Supply Chain Management in Restaurants

Abstract

Logistics optimization in the restaurant industry is important to improve efficiency and reduce operating costs as in any other sector, in this study presents the application of convolutional neural networks (CNN) in the accurate prediction of dishes, which resulted in a 26% improvement in logistics efficiency in a pilot restaurant. The developed system allows restaurants to identify dishes from images uploaded by users, using a CNN model trained with a dataset that includes images of Peruvian meals to achieve these results, several training experiments were implemented with different numbers of epochs (5, 10, 15 and 20 epochs), determining that 20 epochs offered the highest prediction accuracy where software such as TensorFlow and Keras were used for model building and training. The process included the loading and preprocessing of 5000 images, the use of an optimized convolutional neural network, and the prediction of the saucer based on the probabilistic output of the model. The technical approach involved the use of deep learning libraries and advanced image processing techniques to train the model, ensuring its ability to generalize and correctly predict new data, where the system not only improves saucer identification, but also optimizes supply chain and inventory management, enabling more accurate planning and significant waste reduction. The success of this implementation suggests that convolutional neural networks can be a valuable tool in industrial logistics, offering innovative solutions to complex operational challenges. The 26% improvement in logistics efficiency demonstrates the potential of artificial intelligence to transform traditional industrial processes, bringing tangible benefits in terms of both cost and customer satisfaction.