IJRR

International Journal of Research and Review

| Home | Current Issue | Archive | Instructions to Authors | Journals |

Year: 2026 | Month: April | Volume: 13 | Issue: 4 | Pages: 430-437

DOI: https://doi.org/10.52403/ijrr.20260444

A Comparative Study of Deep Learning Algorithms for Fabric Defect Detection

Surbhi Raiyani1, Megha Patel2

1M.S.C Computer Science(ds/ml), P P Savani University, Surat, India,
2Computer Engineering Department, P P Savani University, Surat, India

Corresponding Author: Surbhi Raiyani

ABSTRACT

Manual inspection is tedious and can make mistakes easily. An efficient and accurate AI model has been created to detect defects in fabric automatically.
Lighting within the factory is highly unorganized; therefore, it was necessary to convert the picture to grayscale to improve the lighting of the images. MobileNetV2, a light model, was used for image processing; however, this model was modified by adding an attention mechanism to enable the AI model to pay attention to certain aspects. This will ensure that the AI model will be able to detect defect spots such as snagging, pinholes, and oil marks.
The number of defects within the fabric is much smaller than the number of normal fabrics. Therefore, there was a need to train the AI model in such a way that the model is able to learn through challenging instances until it succeeds.
The result, the proposed model successfully detected the fabric defects 91% of the time. In addition, the efficiency of the proposed model was compared with those of some classical models, and it was found that these classic models had low performances, less than 50-60%.

Keywords: Fabric Defect Detection, Automated Textile, Inspection, Textile Quality Control, Smart Manufacturing, Deep Learning, MobileNetV2, Attention Mechanism

[PDF Full Text]