Design of Inventory Perception and Intelligent Regulation System for Live E-Commerce Based on Deep Image Recognition

Abstract

The fast progress of deep learning and computer vision technology is providing new ways to solve the problem of managing inventory in live e-commerce. As the live e-commerce industry grows, the traditional inventory management model struggles to meet the demands of high concurrency and real-time transactions. The paper puts forward a live e-commerce intelligent inventory management system. This system is based on the improved YOLOv5 model. It is a smart inventory management system for live e-commerce. It combines edge computing technology. This technology enables real-time product recognition and accurate inventory management. The system is designed using a microservice architecture, integrating the GhostNet lightweight network and the SE attention mechanism to significantly improve recognition accuracy and processing efficiency. Using model quantisation and knowledge distillation techniques, the system can overcome the problems caused by complex lighting and background interference. It can also achieve low-latency, highly reliable inventory monitoring and provide intelligent solutions for live e-commerce. Experimental results demonstrate the system’s effectiveness in terms of recognition accuracy, response speed, and resource utilisation, offering a highly efficient and dependable inventory management solution for the live e-commerce sector.