DESIGN AND IMPLEMENTATION OF DATA PROCESSING SUBSYSTEM IN IN-STORE TRAFFIC MONITORING SYSTEM

• Main Author: Imam Maulana, Dalta
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/42130
• Institution: Institut Teknologi Bandung

In-Store Traffic Monitoring System is a system that can provide real-time and easily accessible information such as store density map, number of customers, and store peak hour to retailers. The system uses artificial intelligence to identify customer behavior inside the store. In this system, store image captured by IP camera will be processed using Single Shot Detector – MobileNet v2 object detector and Euclidean distance-based object tracker to determine customer’s position inside the store. Other than that Gaussian kernel density estimation is used to generate store density map from customer’s position data. As a result, the data processing subsystem can produce store density map, number of customers, and store peak hour data with an average data processing speed of 10 frames per second. Futhermore, the system can also send all of the data to server and store the data in local storage if the data cannot be sent to server.