Skip to content

Yohangala/Fabric_Defect_Detection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 

Repository files navigation

🧵 Fabric Defect Detection

An automated fabric defect detection system comparing three deep learning architectures on the TILDA (Tianchi Alibaba) textile defect dataset.


📌 Problem Statement

Manual quality inspection in textile manufacturing is slow and inconsistent. This project trains and benchmarks object detection / classification models to automatically identify defects in fabric images.


🏗️ Models Compared

Model Task Approach
Swin Transformer Classification Hierarchical vision transformer with shifted windows
Faster R-CNN Object Detection Two-stage region proposal network
EfficientNet Classification Compound scaling of depth, width & resolution

📂 Dataset

  • TILDA Textile Dataset — sourced from Alibaba Tianchi competition
  • Contains fabric images labeled with defect regions across multiple defect categories
  • Preprocessing: resizing, normalization, augmentation (flips, rotations)

📁 Repository Structure

Fabric_Defect_Detection/
├── swin_transformer.ipynb   # Swin Transformer training & evaluation
├── faster rcnn.ipynb        # Faster R-CNN training & evaluation
├── efficientnet.ipynb       # EfficientNet training & evaluation
└── README.md

⚙️ Setup

pip install torch torchvision timm albumentations

🔬 Methodology

  1. Data Loading — Load TILDA images and defect annotations
  2. Preprocessing — Resize, normalize, and augment images
  3. Model Training — Train each architecture independently
  4. Evaluation — Compare mAP / accuracy across models
  5. Analysis — Identify best-performing architecture per defect category

📊 Results

Model Metric Score
Swin Transformer Accuracy
Faster R-CNN mAP
EfficientNet Accuracy

Results to be updated after full training runs.


🔗 References


👤 Author

Yohan Gala — B.Tech Computer Engineering, KJSIT Mumbai
GitHub

About

Fabric defect detection using Swin Transformer, Faster R-CNN & EfficientNet on the TILDA textile dataset.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors