Learning and Features based Customer Guidance Function for Application of Smart Autonomous Mover

With the development of deep-learning technologies., the combined use of artificial intelligence and embedded platforms has become increasingly prevalent., and smart autonomous mover is a critical technology. How customers in a crowd can be identified has become a major issue. In this study, we employ deep learning and machine learning technologies to distinguish customers from noncustomers on an embedded platform. The customer guidance function for autonomous mover has two stages, including pre-registration and real-time customer tracking mode. The color space technology is used to compare the registered data of customer's clothes with data acquired by the autonomous mover. Furthermore, gender recognition and face recognition are performed through a classifier. After face, clothes, and gender recognition are conducted, the system identifies the target as a customer or noncustomer. By the proposed methodology, the customer tracking performance for accuracy, precision, recall, and Fl-score is 0.924, 0.999, 0.850, and 0.918, respectively.