Fig 1.
Overall framework of the proposed sealing defect detection system for pharmaceutical plastic bags.
Fig 2.
Thermal imaging comparison between normal and defective sealing regions.
(a) represents a defective sample, and (b) represents a normal sample.
Fig 3.
Collection of real sealing defect samples in thermal imaging.
Fig 4.
Comparison of defect contour fitting results before and after polynomial fitting.
Fig 5.
Comparison between real defects and physics-synthesized defects in thermal imaging.
Fig 6.
Architecture of the proposed dual-branch temporal network for sealing defect detection.
Table 1.
Performance comparison of different networks.
Fig 7.
Training and testing of several networks.
Table 2.
Generalization performance comparison on multiple datasets.