Automated Analysis of Packaging Seal Failures Using Computer Vision Techniques

This work presents a computer vision–based method for the automated detection of sealing failures in carton packaging. In the juice industry, manual inspection of package flaps is destructive, time-consuming, and prone to human error, while defective sealing can lead to product leakage and economic losses. To address these challenges, we propose a two-stage approach: first, images of package flaps are preprocessed to accurately detect and align the flaps and detect the adhesive region; second, the adhesive region is segmented and classified into four categories: Aluminum, Background, Cardboard, and Cardboard-Aluminum using convolutional neural networks applied to image patches. Regions where aluminum appears in the same position on both flaps are identified as sealing failures. Experimental evaluation demonstrates that the method reliably detects defects, reducing reliance on manual inspection and providing an effective solution for automated quality control in industrial production.