基于计算机视觉的智能零售方案

Abstract

In today's retail environment, the drawbacks of the original retail method have been revealed. The main drawbacks are the consumption of a lot of manpower, material resources, store management, financial statistics, preferential programs, event promotions, etc. This project uses information technology to realize unmanned retail, which can solve the traditional manual cash register method. The solution is modular and each module is highly portable. Different combinations of different modules can be adapted to different scenarios. This system mainly uses MTCNN neural network for face recognition, FairMot algorithm for target tracking, RetinaFace algorithm for face recognition and settlement, and data transmission through Java server and Golang server. This system saves a lot of manpower and material resources and has a broad market scenario.