Supermarket commodity identification using convolutional neural networks

In recent years, with the rapid development of deep learning, it has achieved great success in the field of image recognition. In this paper, we applied the convolution neural network (CNN) on supermarket commodity identification, contributing to the study of supermarket commodity identification. Different from the QR code identification of supermarket commodity, our work applied the CNN using the collected images of commodity as input. This method has the characteristics of fast and non-contact. In this paper, we mainly did the following works: 1. Collected a small dataset of supermarket goods. 2. Built Different convolutional neural network frameworks in caffe and trained the dataset using the built networks. 3. Improved train methods by finetuning the trained model.