Papers
Topics
Authors
Recent
Search
2000 character limit reached

YOLO algorithm with hybrid attention feature pyramid network for solder joint defect detection

Published 2 Jan 2024 in cs.CV | (2401.01214v1)

Abstract: Traditional manual detection for solder joint defect is no longer applied during industrial production due to low efficiency, inconsistent evaluation, high cost and lack of real-time data. A new approach has been proposed to address the issues of low accuracy, high false detection rates and computational cost of solder joint defect detection in surface mount technology of industrial scenarios. The proposed solution is a hybrid attention mechanism designed specifically for the solder joint defect detection algorithm to improve quality control in the manufacturing process by increasing the accuracy while reducing the computational cost. The hybrid attention mechanism comprises a proposed enhanced multi-head self-attention and coordinate attention mechanisms increase the ability of attention networks to perceive contextual information and enhances the utilization range of network features. The coordinate attention mechanism enhances the connection between different channels and reduces location information loss. The hybrid attention mechanism enhances the capability of the network to perceive long-distance position information and learn local features. The improved algorithm model has good detection ability for solder joint defect detection, with mAP reaching 91.5%, 4.3% higher than the You Only Look Once version 5 algorithm and better than other comparative algorithms. Compared to other versions, mean Average Precision, Precision, Recall, and Frame per Seconds indicators have also improved. The improvement of detection accuracy can be achieved while meeting real-time detection requirements.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (31)
  1. Smt solder joint inspection via a novel cascaded convolutional neural network. IEEE Transactions on Components, Packaging and Manufacturing Technology, 8(4):670–677, 2018. doi: 10.1109/TCPMT.2018.2789453.
  2. Ic solder joints inspection via an optimized statistical modeling method. IEEE Transactions on Components, Packaging and Manufacturing Technology, 12(2):341–348, 2021. doi: 10.1109/TCPMT.2021.3136720.
  3. Smd led chips defect detection using a yolov3-dense model. Advanced engineering informatics, 47:101255, 2021. doi: 10.1016/j.aei.2021.101255.
  4. An image is worth 16x16 words: Transformers for image recognition at scale. Proceedings of the IEEE conference on computer vision and pattern recognition, 2021. doi: 10.48550/arXiv.2010.11929.
  5. Yolo v4 for advanced traffic sign recognition with synthetic training data generated by various gan. IEEE Access, 9:97228–97242, 2021. doi: 10.1109/ACCESS.2021.3094201.
  6. Towards automatic optical inspection of soldering defects. In 2018 International Conference on Cyberworlds (CW), pages 375–382. IEEE, 2018. doi: 10.1109/CW.2018.00074.
  7. Yolo-mbbi: Pcb surface defect detection method based on enhanced yolov5. Electronics, 12(13):2821, 2023. doi: 10.3390/electronics12132821.
  8. Jocher G. Yolov5. 2022. doi: 10.5281/zenodo.7347926 .
  9. Jocher G. Yolov8. 2023. doi: https://github.com/ultralytics/ultralytics .
  10. An improved yolov5 algorithm for wood defect detection based on attention. IEEE Access, 2023. doi: 10.1109/ACCESS.2023.3293864.
  11. Squeeze-and-excitation networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 7132–7141, 2018. doi: 10.48550/arXiv.1709.01507.
  12. Coordinate attention for efficient mobile network design. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 13713–13722, 2021. doi: 10.48550/arXiv.2103.02907.
  13. Feature pyramid networks for object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2117–2125, 2017. doi: 10.48550/arXiv.1612.03144.
  14. Solder joint defect inspection method based on convnext-yolox. IEEE Transactions on Components, Packaging and Manufacturing Technology, 12(11):1890–1898, 2022. doi: 10.1109/TCPMT.2022.3224997.
  15. Learning spatial fusion for single-shot object detection. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019. doi: 10.48550/arXiv.1911.09516.
  16. Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF international conference on computer vision, pages 10012–10022, 2021. doi: 10.48550/arXiv.2103.14030.
  17. Path aggregation network for instance segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 8759–8768, 2018. doi: 10.1109/CVPR.2018.00913.
  18. Solar cell surface defect detection based on optimized yolov5. IEEE Access, 2023. doi: 10.1109/ACCESS.2023.3294344.
  19. Lff-yolo: A yolo algorithm with lightweight feature fusion network for multi-scale defect detection. IEEE Access, 10:130339–130349, 2022. doi: 10.1109/ACCESS.2022.3227205.
  20. Centralized feature pyramid for object detection. IEEE Transactions on Image Processing, 2023. doi: 10.1109/TIP.2023.3297408.
  21. Faster r-cnn: Towards real-time object detection with region proposal networks. Advances in neural information processing systems, 28, 2015. doi: 10.1109/TPAMI.2016.2577031.
  22. Efficientdet: Scalable and efficient object detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10781–10790, 2020. doi: 10.1109/CVPR42600.2020.01079.
  23. Yolov7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 7464–7475, 2023. doi: 0.48550/arXiv.2207.02696.
  24. Cbam: Convolutional block attention module. In Proceedings of the European conference on computer vision (ECCV), pages 3–19, 2018. doi: 10.48550/arXiv.1807.06521.
  25. Eca-net: Efficient channel attention for deep convolutional neural networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 11534–11542, 2020. doi: 10.48550/arXiv.1910.03151.
  26. Solder joint inspection using eigensolder features. Soldering & Surface Mount Technology, 30(4):227–232, 2018. doi: 10.1108/SSMT-12-2017-0042.
  27. Surface defects detection and identification of lithium battery pole piece based on multi-feature fusion and pso-svm. Ieee Access, 9:85232–85239, 2021. doi: 10.1109/ACCESS.2021.3067641.
  28. Ic solder joint inspection based on an adaptive-template method. IEEE Transactions on Components, Packaging and Manufacturing Technology, 8(6):1121–1127, 2018. doi: 10.1109/TCPMT.2018.2812815.
  29. Application of improved yolov5 model for strip surface defect detection. In 2022 Global Reliability and Prognostics and Health Management (PHM-Yantai), pages 1–5. IEEE, 2022. doi: 10.1109/PHM-Yantai55411.2022.9942194.
  30. Tph-yolov5: Improved yolov5 based on transformer prediction head for object detection on drone-captured scenarios. In Proceedings of the IEEE/CVF international conference on computer vision, pages 2778–2788, 2021. doi: 10.48550/arXiv.2108.11539.
  31. Solder joint defect detection in the connectors using improved faster-rcnn algorithm. Applied Sciences, 11(2):576, 2021. doi: 10.3390/app11020576.
Citations (3)
View on Semantic Scholar

Summary

  • The paper introduces a Hybrid Attention Feature Pyramid Network integrated with YOLOv5, significantly improving context-aware defect detection.
  • It leverages enhanced multi-head self-attention and coordinate attention to retain spatial features, achieving a 91.5% mAP.
  • The proposed model outperforms existing YOLO versions by 4.3%, balancing high precision with real-time processing capabilities.

YOLO Algorithm with Hybrid Attention Feature Pyramid Network for Solder Joint Defect Detection

Introduction

The paper offers an innovative approach that enhances the detection of solder joint defects in surface mount technology (SMT). The traditional manual inspection methods are inefficient, costly, and lack real-time capabilities. This study addresses the challenges of low detection accuracy, high computational costs, and false detection rates by introducing a Hybrid Attention Feature Pyramid Network (HAFPN) integrated with the YOLOv5 framework. The proposed algorithm demonstrates improvements in context-awareness and feature utilization, which significantly increase the detection accuracy while remaining computationally feasible for industrial applications.

Methodology

Overview of the Proposed Approach

The study builds upon existing FPN strategies, integrating a novel Hybrid Attention Mechanism (HAM) composed of enhanced multi-head self-attention (EMSA) and coordinate attention (CA). EMSA improves the contextual perception of the network, while CA resolves the spatial information loss, ensuring better feature retention across different channels. Figure 1

Figure 1

Figure 1: Hybrid Attention Feature Network Architecture. (a)Hybrid Attention Feature Network (HAFPN), (b)Hybrid Attention Mechanism (HAM)

Feature Pyramid Network (FPN) Enhancements

The proposed HAFPN framework combines HAM with YOLOv5's existing architectural components, making it more adept at recognizing and assimilating features from small-sized defects, a known weakness of conventional FPN models. Figure 2

Figure 2

Figure 2: Hybrid Attention Feature Network Architecture. (a)Muti-Head Self Attention, (b)Enhanced Muti-Head Self Attention(EMSA)

Results

Quantitative Evaluation

The experimental results demonstrate the model's superior performance with a mean Average Precision (mAP) of 91.5%, surpassing YOLOv5 by 4.3%. It also improves precision and recall rates for detecting various solder joint defects, evidencing a robust enhancement over existing feature pyramid configurations. Figure 3

Figure 3

Figure 3

Figure 3

Figure 3

Figure 3

Figure 3

Figure 3

Figure 3

Figure 3: Comparison of Heat Map Effects of Different Attention Mechanisms

Comparative Performance

The study compares the improved YOLOv5 detection model against various existing models including YOLOv4, YOLOv5, YOLOv7, YOLOv8, and Faster R-CNN. The results show that the YOLOv5+HAFPN model delivers a balanced compromise between detection accuracy and processing speed, making it suitable for real-time applications. Figure 4

Figure 4

Figure 4: Comparison of defect detection visualization effects between YOLOv5 (a) and YOLOv5+HAFPN (b)

Discussion

The paper provides compelling evidence for integrating hybrid attention mechanisms into feature pyramid networks. By addressing the deficiencies of traditional FPN architectures in processing small defect areas, the proposed methodology achieves superior detection efficacy. Future work could focus on optimizing the model to further reduce computational overhead without sacrificing accuracy, potentially paving the way for its broader adoption in industry settings.

Conclusion

The research introduces a robust enhancement to the YOLOv5 framework through the application of HAFPN, demonstrating substantial improvements in defect detection accuracy, particularly for small solder joint defects. With real-time processing capabilities, this model is valuable for industrial applications, advancing automated quality control processes in manufacturing. Subsequent studies may explore optimizing the model size while maintaining performance metrics, ensuring its practicality in diverse deployment environments.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Sign Up to Generate All Videos Subscribe on YouTube

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Sign Up to Generate

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Generate Now

Collections

Sign up for free to add this paper to one or more collections.

Sign Up