A product identification method for a mixed-reality web shopping system
Hotaka Niwa, Koichi Nagata, Masaya Ohta, Katsumi Yamashita · 2016
On an e-commerce site, users can easily search for a desired product by inputting the name and/or the model number of the product into a web browser. However, they fail to find it if the query is inputted incorrectly or an ambiguous term is used. We have proposed a mixed-reality web shopping system with panoramic images photographed at the aisles of a real store. In this system, the user can move freely around the store and pick the desired product up while viewing the panoramic images. To implement the system, the product in the panoramic image must be recognized automatically. Because there are a huge number of products in the image, it is not practical that all products are recognized manually. In this paper we consider a product identification method for the mixed-reality web shopping system. The convolutional neural network was used to recognize products in this method.