A Proposal for Article Management Method Using Wearable Camera

摘要

Inventory management is an essential maintenance procedure in various fields such as factories, warehouses, and stores. However, efficiently managing the various types of management targets is a difficult task, and various management methods have been implemented in practice. In this study, I focus that an inventory is changed by the incoming and outgoing activities of workers. And, I propose a deep learning method that automatically acquires a target inventory from video images of the workers activities. The videos are captured by wearable cameras attached to the workers. To improve the distinction accuracy, different deep learning models are trained for different situations, and multiple consecutive images of one target extracted from the video are used. By this method, it is expected inventories can be simply managed as workers move around the inventory shelves.