Product identification of wine types in shopping centers through deep learning for visual assistance. The ability of convolutional neural networks to detect the difference between one product or another of the same class on shopping center shelves has been tested. In this study, we evaluate the performance of a model based on Residual Network (ResNet), a pre-trained model of a neural network, which only modified the output layer of the network, freezing its weights to train the output layer with the images of products from shopping centers taken directly from the displays, adapting the images to the model to the parameters required by the aforementioned model. The code used labels the captured images in two different classes and is subsequently tested to see if it finds the difference between one class or another. During this work, six experiments are carried out that challenge the network, starting from recognizing whether it is a wine or another object, and ending up knowing if it is capable of detecting a specific wine among other different ones.