Scholarlink

In supermarkets, product recognition for billing and stock management is labor-intensive and inefficient. To address these issues, a real-time commodity identification system using computer vision and deep learning is proposed. The smart shopping system improves consumer experience with automatic cart billing, reducing checkout time and stock update alerts. Custom dataset of retail products are used for training YOLOv8, with fixed multiple cameras to capture different views to identify hidden objects and objects of different scales using instance segmentation. We achieve a maximum F1 score of 0.978 across all products with the least training time compared to other detection models.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Commodity Identification Using Deep Learning in Smart Shoppping System

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • P. Dhevanathan,
  • S. Mary Saira Bhanu

摘要

In supermarkets, product recognition for billing and stock management is labor-intensive and inefficient. To address these issues, a real-time commodity identification system using computer vision and deep learning is proposed. The smart shopping system improves consumer experience with automatic cart billing, reducing checkout time and stock update alerts. Custom dataset of retail products are used for training YOLOv8, with fixed multiple cameras to capture different views to identify hidden objects and objects of different scales using instance segmentation. We achieve a maximum F1 score of 0.978 across all products with the least training time compared to other detection models.