Lightweight CNN-Based Framework for Industrial Surface Defect Classification
DOI:
https://doi.org/10.25195/ijci.v52i1.769Keywords:
ndustrial Surface Defect Classification, Convolutional Neural Networks (CNN), Deep Learning, , Hyperparameter Tuning, Grad-CAMAbstract
Industrial surface defect classification is an important part of automated quality inspection systems. For these systems to work, they need to be able to detect surface defects accurately and efficiently to improve product reliability and reduce manufacturing costs. Traditional manual inspection methods are often time-consuming, subjective, and not suitable for fast-paced industrial environments. This study proposes a lightweight convolutional neural network (CNN)-based system for classifying industrial surface defects. The model was made to work well for classification and still be fast enough for real-world use. Keras Tuner was used in the proposed method to find the best hyperparameters. The proposed model was evaluated on the Northeastern University (NEU) Surface Defect Database using 5-fold stratified cross-validation. The results obtained from the experiments are promising since the system yields stable performance on all folds with accuracy scores of 99.72%, 99.17%, 100.00%, 99.17%, and 98.61% respectively. The mean accuracy score is calculated as 99.33%. Also, Grad-CAM visualization revealed that the network focuses on defective regions when processing an input image which supports the reliability and interpretability of the classification process.
Downloads
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Iraqi Journal for Computers and Informatics
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
IJCI applies the Creative Commons Attribution (CC BY) license to articles. The author of the submitted paper for publication by IJCI has the CC BY license. Under this Open Access license, the author gives an agreement to any author to reuse the article in whole or part for any purpose, even for commercial purposes. Anyone may copy, distribute, or reuse the content as long as the author and source are properly cited. This facility helps in re-use and ensures that journal content is available for the needs of research.
If the manuscript contains photos, images, figures, tables, audio files, videos, etc., that the author or the co-authors do not own, IJCI will require the author to provide the journal with proof that the owner of that content has given the author written permission to use it, and the owner has approved that the CC BY license being applied to content. IJCI provides a form that the author can use to ask for permission from the owner. If the author does not have owner permission, IJCI will ask the author to remove that content and/or replace it with other content that the author owns or has such permission to use.
Many authors assume that if they previously published a paper through another publisher, they have the right to reuse that content in their PLOS paper, but that is not necessarily the case – it depends on the license that covers the other paper. The author must ascertain the rights he/she has of a specific license (a license that enables the author to use the content). The author must obtain written permission from the publisher to use the content in the IJCI paper. The author should not include any content in her/his IJCI paper without having the right to use it, and always give proper attribution.
The accompanying submitted data should be stated with licensing policies, the policies should not be more restrictive than CC BY.
IJCI has the right to remove photos, captures, images, figures, tables, illustrations, audio, and video files, from a paper before or after publication, if these contents were included in the author's paper without permission from the owner of the content.
