Study on Real-Time Terahertz Imaging and Object Detection System for Soft Foreign Matter in Food Using YOLOv8 with Attention Mechanism

Howon Yoon, Gukjin Son, Seungeon Song, Jonghun Lee, Youngduk Kim

Abstract


In this paper, we propose a real-time detection system for identifying soft foreign matter in food, leveraging terahertz imaging technology combined with the YOLOv8 object detection algorithm. Terahertz waves, known for their ability to penetrate non-metallic and non-polar materials, enable nondestructive inspection of internal structures. Focusing on seaweed as the primary food product, we developed a specialized dataset and conducted experiments to detect foreign matter. By integrating several attention mechanism modules, including ECA,
CBAM and others, into the YOLOv8 framework, we achieved enhancements in both detection accuracy and processing speed. Our results highlight the system's effectiveness in identifying soft
foreign matter, and we anticipate that this approach will significantly improve food safety and quality control within the industry. This study provides significant contributions to the
application of terahertz imaging and deep learning in the domain of food safety, expecting a positive impact on future food management systems.


Keywords


Terahertz Imaging; Attention Mechanisms; Object Detection; Deep Learning; Food Safety

Full Text:

PDF

Refbacks

  • There are currently no refbacks.