Fig 1.
Improved DCFE-YOLOv8 algorithm.
Fig 2.
Overall structure of DSConv.
Fig 3.
Improvements in the backbone network’s C2f module.
Fig 4.
Structure of CPCA.
Fig 5.
FE_Block.
Fig 6.
PConv structure.
Fig 7.
Overall structure of EMA attention mechanism.
Fig 8.
FE_C2f.
Fig 9.
Fabric defect samples.
Table 1.
Experimental environment configuratio.
Table 2.
Comparison of different attention mechanisms.
Fig 10.
Comparison of CPCA attention mechanism effects.
Table 3.
Ablation experiments of different modules.
Table 4.
Model comparison experiments.
Table 5.
Comparison of mAP@0.5 for different defect categories.
Fig 11.
Verification of the fitting degree between the improved model and the data.
Fig 12.
Before and after improvement mAP@0.5 comparison of indicators.
Fig 13.
Comparison of results for fabric defect detection before and after improvement.