An Enhanced YOLOv4 Model With Self-Dependent Attentive Fusion and Component Randomized Mosaic Augmentation for Metal Surface Defect Detection
Metal surface quality control is significant in the production line of metal products. Detecting metal surface defects is challenging due to the various types and morphological patterns. Recent advances have witnessed deep learning-based automated optical inspection systems as a promising solution....
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9870791/ |
| _version_ | 1798000644767875072 |
|---|---|
| author | Chenglong Wang Ziran Zhou Zhiming Chen |
| author_facet | Chenglong Wang Ziran Zhou Zhiming Chen |
| author_sort | Chenglong Wang |
| collection | DOAJ |
| description | Metal surface quality control is significant in the production line of metal products. Detecting metal surface defects is challenging due to the various types and morphological patterns. Recent advances have witnessed deep learning-based automated optical inspection systems as a promising solution. This paper presents an enhanced YOLOv4 model for metal surface defect detection (MSDD). Specifically, we integrate three boosting components into the original YOLOv4, including 1) a self-dependent attentive fusion (SAF) block, placed within the model neck, to enhance inter-path and cross-layer feature fusion, 2) a component randomized Mosaic augmentation (CRMA) scheme to strategically discourage an over-transformed image to participate in training, and 3) a perturbation agnostic (PA) label smoothing method to keep the model from making over-confident predictions and thus act as a means of regularization. The proposed method has been validated on a self-developed MSDD dataset. It is shown that each boosting component can lead to an impressive mAP gain, and the final model outperforms the baselines, namely, Faster R-CNN, YOLOv4, YOLOv5, and YOLOX, by 7.85%, 6.51%, 3.76%, and 3.57%, respectively. |
| first_indexed | 2024-04-11T11:23:37Z |
| format | Article |
| id | doaj.art-4bb5c1c5014f4dd6b9a0cb4ba947b2a5 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-11T11:23:37Z |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-4bb5c1c5014f4dd6b9a0cb4ba947b2a52022-12-22T04:26:23ZengIEEEIEEE Access2169-35362022-01-0110977589776610.1109/ACCESS.2022.32031989870791An Enhanced YOLOv4 Model With Self-Dependent Attentive Fusion and Component Randomized Mosaic Augmentation for Metal Surface Defect DetectionChenglong Wang0Ziran Zhou1Zhiming Chen2School of Electronic Information and Electrical Engineering, Huizhou University, Huizhou, ChinaSchool of Electronic Information and Electrical Engineering, Huizhou University, Huizhou, ChinaSchool of Electronic Information and Electrical Engineering, Huizhou University, Huizhou, ChinaMetal surface quality control is significant in the production line of metal products. Detecting metal surface defects is challenging due to the various types and morphological patterns. Recent advances have witnessed deep learning-based automated optical inspection systems as a promising solution. This paper presents an enhanced YOLOv4 model for metal surface defect detection (MSDD). Specifically, we integrate three boosting components into the original YOLOv4, including 1) a self-dependent attentive fusion (SAF) block, placed within the model neck, to enhance inter-path and cross-layer feature fusion, 2) a component randomized Mosaic augmentation (CRMA) scheme to strategically discourage an over-transformed image to participate in training, and 3) a perturbation agnostic (PA) label smoothing method to keep the model from making over-confident predictions and thus act as a means of regularization. The proposed method has been validated on a self-developed MSDD dataset. It is shown that each boosting component can lead to an impressive mAP gain, and the final model outperforms the baselines, namely, Faster R-CNN, YOLOv4, YOLOv5, and YOLOX, by 7.85%, 6.51%, 3.76%, and 3.57%, respectively.https://ieeexplore.ieee.org/document/9870791/Metal surface defect detectiondeep learningYOLOself-dependent attentive fusioncomponent randomized Mosaic augmentationlabel smoothing |
| spellingShingle | Chenglong Wang Ziran Zhou Zhiming Chen An Enhanced YOLOv4 Model With Self-Dependent Attentive Fusion and Component Randomized Mosaic Augmentation for Metal Surface Defect Detection IEEE Access Metal surface defect detection deep learning YOLO self-dependent attentive fusion component randomized Mosaic augmentation label smoothing |
| title | An Enhanced YOLOv4 Model With Self-Dependent Attentive Fusion and Component Randomized Mosaic Augmentation for Metal Surface Defect Detection |
| title_full | An Enhanced YOLOv4 Model With Self-Dependent Attentive Fusion and Component Randomized Mosaic Augmentation for Metal Surface Defect Detection |
| title_fullStr | An Enhanced YOLOv4 Model With Self-Dependent Attentive Fusion and Component Randomized Mosaic Augmentation for Metal Surface Defect Detection |
| title_full_unstemmed | An Enhanced YOLOv4 Model With Self-Dependent Attentive Fusion and Component Randomized Mosaic Augmentation for Metal Surface Defect Detection |
| title_short | An Enhanced YOLOv4 Model With Self-Dependent Attentive Fusion and Component Randomized Mosaic Augmentation for Metal Surface Defect Detection |
| title_sort | enhanced yolov4 model with self dependent attentive fusion and component randomized mosaic augmentation for metal surface defect detection |
| topic | Metal surface defect detection deep learning YOLO self-dependent attentive fusion component randomized Mosaic augmentation label smoothing |
| url | https://ieeexplore.ieee.org/document/9870791/ |
| work_keys_str_mv | AT chenglongwang anenhancedyolov4modelwithselfdependentattentivefusionandcomponentrandomizedmosaicaugmentationformetalsurfacedefectdetection AT ziranzhou anenhancedyolov4modelwithselfdependentattentivefusionandcomponentrandomizedmosaicaugmentationformetalsurfacedefectdetection AT zhimingchen anenhancedyolov4modelwithselfdependentattentivefusionandcomponentrandomizedmosaicaugmentationformetalsurfacedefectdetection AT chenglongwang enhancedyolov4modelwithselfdependentattentivefusionandcomponentrandomizedmosaicaugmentationformetalsurfacedefectdetection AT ziranzhou enhancedyolov4modelwithselfdependentattentivefusionandcomponentrandomizedmosaicaugmentationformetalsurfacedefectdetection AT zhimingchen enhancedyolov4modelwithselfdependentattentivefusionandcomponentrandomizedmosaicaugmentationformetalsurfacedefectdetection |