Current unmanned retail technologies can be roughly divided into the following three schools:

1. Artificial intelligence school: represented by Amazon Go, Ali Tao Coffee, and Take Go, mainly using machine vision, deep learning algorithms, sensor fusion technology, convolutional neural networks, biometrics and other technologies.

2. Internet of Things school: represented by Bingo Box, mainly using RFID tag technology.

3. Internet school: represented by Convenience Bee and Xiaoe Micro Store, mainly using QR codes to complete the identification of goods.

From the current trial operation and commercial unmanned retail stores, the current mainstream technology types of unmanned retail are mainly the following four types of technology:

The first is the QR code scanning technology commonly used in traditional supermarkets. This technology is relatively simple, but the consumer scanning and payment experience is not intelligent enough.

The second is radio frequency identification (RFID), an electronic tag that uses radio frequency communication to achieve contactless automatic identification, and can complete information input and processing without direct contact, laser scanning, or human intervention. Many unmanned convenience stores currently in operation use this technology, such as Bingo Box.

The third is visual recognition technology based on convolutional neural networks, which teaches computers to recognize products through deep learning. When consumers purchase by themselves, the computer uses the store’s cameras to identify the category and price of the products, and the payment can be automatically deducted without the need for consumers to check out manually.

The fourth is to add video recognition and sensor hybrid technology on the basis of visual recognition, adding more algorithms and face recognition technology. Amazon Go, which is currently in the testing phase, uses this technology, but its disadvantage is that it is too expensive.