Title: A defect detection algorithm for ham sausage packaging

Authors: Mengxia Zhang; Haixing Wang; Zhuoran Zhang; Qunpo Liu

Addresses: School of Electrical Engineering and Automation, Henan Polytechnic University, Jiaozuo, Henan, China ' School of Electrical Engineering and Automation, Henan Polytechnic University, Jiaozuo, Henan, China ' School of Electrical Engineering and Automation, Henan Polytechnic University, Jiaozuo, Henan, China ' School of Electrical Engineering and Automation, Henan Polytechnic University, Jiaozuo, Henan, China

Abstract: This article constructs an improved SSD model for detecting surface defects in ham sausage packaging. In response to the problem of inaccurate position information and low resolution after multiple convolution operations on deep features in SSD algorithm, this paper constructs a feature pyramid and multi-scale residual fusion structure. Through feature fusion, the feature extraction performance of the model is effectively improved, enabling the model to extract effective feature layers containing more defect information. In response to the problem of complex surface patterns in ham sausage packaging leading to inaccurate defect localisation accuracy, this article utilises an efficient channel attention module to enhance the model's attention to surface defects in ham sausage packaging. Through experiments, it has been shown that the detection accuracy of ham sausage packaging defects has been improved from 95.84 to 98.30%, and the detection speed has been increased from 33.55 FPS to 33.76 FPS.

Keywords: defect detection; characteristic pyramid; multi-scale residual fusion; efficient channel attention.

DOI: 10.1504/IJCCPS.2024.145820

International Journal of Cybernetics and Cyber-Physical Systems, 2024 Vol.1 No.4, pp.321 - 331

Received: 09 Oct 2023
Accepted: 04 Nov 2023
Published online: 25 Apr 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article