CoChat
Sign Up
Cited by 345Industrial Vision Systems and Defect DetectionNon-Destructive Testing TechniquesAdvanced Neural Network Applications

Steel defect classification with Max-Pooling Convolutional Neural Networks

Open access
Masci, Jonathan, Meier, Ueli, Ciresan, Dan, Schmidhuber, Jurgen, Fricout, Gabriel·2012·The 2012 International Joint Conference on Neural Networks (IJCNN)
DOI 10.1109/ijcnn.2012.6252468
Open this paper in CoChat — free account

Abstract

We present a Max-Pooling Convolutional Neural Network approach for supervised steel defect classification. On a classification task with 7 defects, collected from a real production line, an error rate of 7% is obtained. Compared to SVM classifiers trained on commonly used feature descriptors our best net performs at least two times better. Not only we do obtain much better results, but the proposed method also works directly on raw pixel intensities of detected and segmented steel defects, avoiding further time consuming and hard to optimize ad-hoc preprocessing.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1109/ijcnn.2012.6252468

Cites

30 of 30 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Steel surfaces exhibit many types of defects, and NN-BP has been extensively used to classify such defects.

    “Steel surfaces exhibit many types of defects, and NNBP has been extensively used to classify such defects [31,36,54,59,63,72,78,80].”

    From Review of vision-based steel surface inspection systems (2014) · cites this paper as [80]

    The Introduction describes steel-surface defect classification using a max-pooling convolutional neural network, and the Training Procedure says the network uses back-propagation. That supports an example of NN-BP applied to the task, but not the claim that it has been extensively used; the Introduction instead says there is not much literature on steel defect detection. The cited passages do not establish how many types of defect occur. (LLM, conf=0.84)

    Experimental — accuracy not yet measured

  • With 7% error rate, MPCNN performed much better than SVM for seven defects in cold strips.

    “With 7% error rate, MPCNN performed much better than SVM for seven defects in cold strips.”

    From Review of vision-based steel surface inspection systems (2014) · cites this paper as [80]

    The Introduction (section I, page unavailable) says the paper applies MPCNN to steel-defect classification and compares it with classifiers using conventional features. Searches for “7%,” “SVM,” seven defect classes, and experimental results did not retrieve a passage giving the claimed error rate or SVM comparison. The available passages do not reach the results needed to verify the claim. (LLM, conf=0.68)

    Experimental — accuracy not yet measured

  • For deep learning detection algorithms, detection accuracy increases with data volume and real-time performance is better than traditional algorithms.

    “For deep learning detection algorithms (such as M-Pooling CNN [17], Improved YOLO [19]), their detection accuracy increases with data volume and their real-time performance is better than that of the traditional algorithms.”

    From An Improved VGG19 Transfer Learning Strip Steel Surface Defect Recognition Deep Neural Network Based on Few Samples and Imbalanced Datasets (2021) · cites this paper as [17]

  • Masci J. achieved 90% accuracy on 7 classes with 150x150 input and 6.2 ms execution time.

    “Masci J. [7] 150x150 6.2 ms 90% 7”

    From Real-time system for automatic cold strip surface defect detection (2019) · cites this paper as [7]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

1 of the 31 references couldn’t be matched to a known paper.

Cited by

Sorted by citations

Related

Authors
Masci, J., Meier, U., Ciresan, D., Schmidhuber, J., & Fricout, G.
Affiliations
University of Applied Sciences and Arts of Southern Switzerland (CH)
ORCID
Masci, Jonathan
Venue
The 2012 International Joint Conference on Neural Networks (IJCNN)
Publisher
IEEE
Published
2012-06
Pages
1-6
ISSN
2161-4407 (ISSN-L)
DOI
10.1109/ijcnn.2012.6252468
OpenAlex
W1981759979
Cited by
345
References
31
Citations by year
2026: 5 · 2025: 28 · 2024: 36 · 2023: 26 · 2022: 39 · 2021: 54 · 2020: 52 · 2019: 47 · 2018: 28 · 2017: 11 · 2016: 8 · 2015: 2 · 2014: 3 · 2013: 5 · 2012: 1
FWCI
8.80
Type
conference paper
Language
en
Open access
Licence
public-domain
Locations
  • International Joint Conference on Neural Networks (IJCNN) (publishedVersion)
  • Zenodo (CERN European Organization for Nuclear Research) (submittedVersion, public-domain)
Topics
  • Physical Sciences › Engineering › Industrial and Manufacturing Engineering › Industrial Vision Systems and Defect Detection
  • Physical Sciences › Engineering › Mechanical Engineering › Non-Destructive Testing Techniques
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
Keywords
max pooling, convolutional neural networks, defect detection, defect segmentation

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review