Lightweight edge-attention network for surface-defect detection of rubber seal rings
Bibliographic record
Abstract
Abstract Detection of surface defects is an essential step to ensure the quality and reliability of rubber seal-ring products. Machine vision inspection methods have received attention from engineers because they are beneficial in terms of the detection speed or detection accuracy. It is important to simultaneously guarantee high speed and high accuracy of defect detection in practice production. Considering this problem, in this paper, we propose a YOLO-based (You Only Look Once) lightweight neural network with edge-attention fusion for defect detection of rubber rings. First, an edge-attention block is integrated into the network to improve the detection accuracy. The edge images are weighted with attention weights and summarized with feature maps in the block. Second, an involution-based feature pyramid structure is proposed to reduce the network size and computational cost by using a new involution operator, instead of a convolution operator, at the bottleneck of the feature pyramid. Comparison of the proposed network with popular YOLO networks was performed based on a rubber-ring-image dataset with 138 images. The network achieved the highest detection accuracy and fastest speed among selected lightweight networks, including YOLOv3-tiny, YOLOv5-s, and YOLOX-s. Moreover, compared with the large and highly accurate YOLOv3-spp network, the network size decreased by 93.45% and the computational cost decreased by 91.46%, while average precision only decreased by 3.41%. It can be concluded that the YOLO-based lightweight neural network reaches a good balance between detection speed and detection accuracy.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".