Automated Surface Quality Control via Deep Visual Inspection in Manufacturing: A ResNet50 Implementation Case Study
Bibliographic record
Abstract
Effective quality assurance in industrial manufacturing hinges on the accurate detection and classification of surface anomalies.Conventional manual visual inspection methods are inherently limited by operator subjectivity and fatigue, leading to inconsistent results and potential errors.To minimize these limitations, automated inspection systems utilizing Artificial Intelligence (AI), specifically Convolutional Neural Networks (CNNs), have been implemented.Among the widely adopted CNN architectures for this application are ResNet50, MobileNetV3, and EfficientNet-B0.The ResNet50 which is characterized by its deep residual learning framework exhibits superior classification accuracy, rendering it particularly suitable for identifying nuanced and complex defects.The second one, MobileNetV3, engineered for low-latency inference on mobile or resource-constrained hardware, offers accelerated processing but compromises slightly on accuracy.The last one, EfficientNet-B0, provides a balanced trade-off between accuracy and computational efficiency, yet its performance is surpassed by ResNet50 when classifying intricate defect patterns.Our findings confirm that ResNet50 demonstrates superior performance, especially for high-fidelity detection tasks involving defect types like cracks, stains, and deformations.Although MobileNetV3 and EfficientNet-B0 serve well in real-time or lightweight system deployments, ResNet50 remains the optimal choice for industrial contexts where maximizing detection accuracy is paramount, owing to its robustness in modeling complex defect characteristics and delivering reliable classifications.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".