All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
Dear authors, we are pleased to verify that you meet the reviewer's valuable feedback to improve your research.
Thank you for considering PeerJ Computer Science and submitting your work.
Kind regards
PCoelho
[# PeerJ Staff Note - this decision was reviewed and approved by Mehmet Cunkas, a PeerJ Section Editor covering this Section #]
It is OK for me.
It is OK for me.
It is OK for me.
Dear authors,
Thanks a lot for your efforts to improve the manuscript.
Nevertheless, some concerns are still remaining that need to be addressed.
Like before, you are advised to critically respond to the remaining comments point by point when preparing a new version of the manuscript and while preparing for the rebuttal letter.
Kind regards,
PCoelho
The research addressed in this paper tackles a critical challenge in industrial applications: balancing detection accuracy and model complexity for PCB defect detection on resource-constrained platforms. This is a highly relevant and practical problem, as lightweight models are essential for real-time deployment in industrial settings where computational resources may be limited. The authors’ focus on optimizing YOLOv8n, a widely used baseline, provides a solid foundation for comparison, and the enhanced PKU-Market-PCB dataset serves as a suitable testbed to validate the proposed method. Overall, the motivation and scope of the work are well-justified.
The experimental results on the enhanced PKU-Market-PCB dataset show promising improvements over YOLOv8n, with significant reductions in parameters, computational cost, and model size, accompanied by a 1.7% increase in accuracy. These metrics effectively validate the model’s efficiency and performance. To further support the claims, it would be beneficial to include comparisons with other state-of-the-art lightweight models (e.g., MobileNet-based detectors, YOLOv5s, or EfficientDet) to contextualize the proposed method’s superiority. Additionally, providing ablation studies for each key component (e.g., C2f_Star, GS_Detect, Inner-MPDIoU, and LAMP) would clarify their individual contributions to the overall performance gains.
The integration of group convolution and adaptive pruning (LAMP) further aligns with the goal of lightweight design. These innovations are logically connected, forming a cohesive strategy to address the trade-off between accuracy and efficiency. However, a more detailed explanation of how C2f_Star specifically enhances the fusion capabilities of the original C2f structure, and why Inner-MPDIoU is more effective for small PCB defects compared to other advanced loss functions (e.g., SIoU, EIoU), would strengthen the novelty.
Dear authors,
You are advised to critically respond to all comments point by point when preparing an updated version of the manuscript and while preparing for the rebuttal letter. Please address all comments/suggestions provided by reviewers, considering that these should be added to the new version of the manuscript.
Kind regards,
PCoelho
The manuscript titled “A lightweight printed circuit board defect detection method based on group convolution and adaptive pruning” presents a well-structured study aimed at addressing the challenge of balancing detection accuracy and model complexity in PCB defect detection on resource-constrained industrial platforms. The authors' approach of leveraging the YOLOv8n framework as a baseline and introducing multiple novel techniques, such as the C2f_Star module, GS_Detect head, Inner-MPDIoU loss function, and LAMP for pruning, is commendable. The paper demonstrates a clear understanding of the existing problems in the field and offers a comprehensive solution with promising experimental results.
The literature review in the introduction section could be more comprehensive. While the authors mention the general trade-off between accuracy and complexity in PCB defect detection, they could further explore and discuss recent advancements in lightweight deep learning models for similar tasks in more detail. This would help to better position the proposed method within the current state-of-the-art research.
The authors have conducted thorough experiments on the augmented PKU - Market - PCB dataset. The comparison with the baseline YOLOv8n model, clearly showing significant reductions in parameters (60%), computational cost (51%), model size (60%), and an absolute improvement of 1.7 percentage points in detection accuracy, strongly validates the effectiveness of the proposed method. The experimental results are well-presented with appropriate statistical measures, making the claims of the paper convincing.
Some of the technical descriptions in the paper lack sufficient detail. For example, when introducing the C2f_Star module, GS_Detect head, and the Inner-MPDIoU loss function, more in-depth explanations of the internal working mechanisms and the rationales behind the design choices could be provided. This would make it easier for readers, especially those who are not familiar with the specific techniques, to understand and reproduce the proposed method.
The paper does not adequately discuss the limitations of the proposed method. There may be scenarios or types of PCB defects that the model does not perform well on. Addressing these potential limitations would give a more complete picture of the method's applicability and help readers to better evaluate the research.
The paper does not adequately discuss the limitations of the proposed method. There may be scenarios or types of PCB defects that the model does not perform well on. Addressing these potential limitations would give a more complete picture of the method's applicability and help readers to better evaluate the research.
Suggestions for Revision
Enhance Related Work: Expand the literature review section to include a more detailed analysis of recent research on lightweight deep learning models for defect detection in general and PCB defect detection in particular. Compare the proposed method more thoroughly with these existing approaches in terms of methodology, performance, and applicability.
Clarify Technical Details: Provide more in-depth technical descriptions of the novel components proposed in the paper. This could include additional diagrams, equations, or step-by-step explanations to clarify the working principles of the C2f_Star module, GS_Detect head, and Inner - MPDIoU loss function. Also, explain the hyperparameter settings and optimization strategies for these components.
Discuss Limitations: Add a section to explicitly discuss the limitations of the proposed method. Analyze potential failure cases, such as specific types of defects, imaging conditions, or industrial environments where the model may not perform optimally. Propose possible future research directions to address these limitations.
This paper presents a well-executed and comprehensive approach to lightweight PCB defect detection. One of its key strengths is the thoughtful integration of the Inner-MPDIoU loss function to compensate for potential performance degradation caused by model simplifications, such as the C2f_Star module, the GS_Detect module, and LAMP model pruning. By effectively addressing the trade-off between model efficiency and detection accuracy, the authors provide a holistic solution suitable for deployment in resource-constrained industrial environments.
Suggestions for Improvement
Figure 12 Clarity: While the heatmaps in Figure 12 offer visual insight into the model's attention regions, the claim regarding missed detections is not directly verifiable to readers. It is recommended to include ground truth (GT) annotations in the figure to make the analysis more transparent and interpretable.
-
Suggestions for Improvement
Figure 10 Annotation: The defect examples presented in Figure 10 would benefit from clearer labeling. Specifically, adding the defect names below each image would improve readability and help readers better understand the model’s performance across different defect categories.
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.