Title: Detection with thermal imaging for packaging bag sealing based on knowledge transfer

Authors: Shaoyu Tang; Lisheng Wei; Rui Wang; Pinggai Zhang

Addresses: School of Electrical Engineering, Anhui Polytechnic University, Anhui, 241000, China ' School of Electrical Engineering, Anhui Polytechnic University, Anhui, 241000, China ' School of Computer Science and Technology, China University of Mining and Technology, Jiangsu, 221116, China ' School of Electronic Engineering, Chaohu University, Anhui, 238000, China

Abstract: To address the problem that most enterprises still use the manual packaging bag seal detection method with low efficiency and poor stability, we propose an automatic detection method, which is based on knowledge transfer, to detect with thermal imaging the packaging bag sealing. Firstly, the thermal image of packaging bag seal is obtained by a thermal imager, random forest (RF) and support vector machine (SVM) are trained by small sample labels, and the two classifiers are fused to build an expert labelling system for labelling unlabelled samples. Then, the enhanced samples are created by combining the predicted samples and the labelled samples, and input into the fine-tuned VGG16 (visual geometry group) for training and testing. Finally, the experiment shows that the prediction accuracy of this method reaches 96.25%, which verifies the effectiveness and feasibility of the proposed method instead of manual detection method.

Keywords: defect detection; expert labelling system; fine-tuned VGG16; knowledge transfer; thermal imaging.

DOI: 10.1504/IJMIC.2024.141677

International Journal of Modelling, Identification and Control, 2024 Vol.45 No.1, pp.58 - 69

Received: 27 Jul 2023
Accepted: 29 Feb 2024
Published online: 30 Sep 2024 *