SimTho/IndustrialTextileDataset

1<p align="center">2  <img width="700" height="400" src="images/LogoITD.png">3</p>4 5## Description6Introduction of new dataset for unsupervised fabric defect detection 7This dataset aims to provide a color dataset with real industrial fabric defect gathered in a visiting machine with several industrial cameras.8It has been designed with the same nomenclature as MVTEC AD dataset (https://www.mvtec.com/company/research/datasets/mvtec-ad) for unsupervised anomaly detection. 9 10<p align="center">11  <img width="700" height="250" src="images/Samples.png">12</p>13 14<div align="center"> 15  16| Type        | Total      | Train(Good) | Test(Good) | Test(Defective)  | Sample | 17| :------:|:-----:|:-----:| :------:|:-----:|-----|18| type1cam1  	| 386 	| 272 	| 28 	| 86 	| <img src="images/type1cam1.png" alt="" width="150"> |19| type2cam2  	| 257 	| 199 	| 19 	| 39 	| <img src="images/type2cam2.png" alt="" width="150">|20| type3cam1  	| 689 	| 588 	| 54 	| 47 	| <img src="images/type3cam1.png" alt="" width="150">|21| type4cam2  	| 229 	| 199 	| 19 	| 11 	| <img src="images/type4cam2.png" alt="" width="150">|22| type5cam2  	| 298 	| 199 	| 19 	| 80 	| <img src="images/type5cam2.png" alt="" width="150">|23| type6cam2  	| 291 	| 199 	| 19 	| 73 	| <img src="images/type6cam2.png" alt="" width="150">|24| type7cam2  	| 917 	| 711 	| 89 	| 117 	| <img src="images/type7cam2.png" alt="" width="150">|25| type8cam1  	| 868 	| 711 	| 89 	| 68 	| <img src="images/type8cam1.png" alt="" width="150">|26| type9cam2 	| 856 	| 721 	| 86 	| 49 	| <img src="images/type9cam2.png" alt="" width="150">|27| type10cam2 	| 871 	| 717 	| 90 	| 64 	| <img src="images/type10cam2.png" alt="" width="150">|28 29</div>30 31## Download 32 33The dataset can be downloaded in google drive with this link : [LINK](https://drive.google.com/drive/folders/1orrMLs0FH4KgEm0vIsneeX3qsvILMh6L?usp=sharing) 34 35 36 37## Utilisation38This dataset is designed for unsupervised anomaly detection task but can also be used for domain-generalization approach.39The nomenclature is designed as : 40<p align="center">41  <img width="550" height="350" src="images/Nomenclature2.png">42</p>43 44- category/45  - train/46    - good/47      - img1.png48      - ...49  - test/50    - anomaly/51      - img1.png52      - ...53    - good/54      - img1.png55      - ...56 57As in any unsupervised training, train data are defect-free. Defective samples are only in the test set.58 59## Exemples60 61Exemple of defect segmentation obtained with our knowledge distillation-based method62<p align="center">63  <img width="700" height="250" src="images/DefectITDB.png">64</p>65 66 67## Documentation68 69List of articles related to the subject of textile defect detection70 71- **MixedTeacher : Knowledge Distillation for fast inference textural anomaly detection** (https://arxiv.org/abs/2306.09859)72- **FABLE : Fabric Anomaly Detection Automation Process** (https://arxiv.org/abs/2306.10089)73- **Exploring Dual Model Knowledge Distillation for Anomaly Detection** (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4493018)74- **Distillation-based fabric anomaly detection** (https://journals.sagepub.com/doi/abs/10.1177/00405175231206820)(https://arxiv.org/abs/2401.02287)75## Auteurs76 77- Simon Thomine <sup>1</sup>, PhD student - [@SimonThomine](https://github.com/SimonThomine) - simon.thomine@utt.fr78- Hichem Snoussi <sup>1</sup>, Full Professor79 80<sup>1</sup> University of Technology of Troyes, France81 82## Citation83If you use this dataset, please cite84```85@inproceedings{Thomine_2023_Knowledge,86    author    = {Thomine, Simon and Snoussi, Hichem},87    title     = {Distillation-based fabric anomaly detection},88    booktitle = {Textile Research Journal},89    month     = {August},90    year      = {2023}91}92```93 94## Licence95 96 97This project is under the MIT license [MIT](https://opensource.org/licenses/MIT).98---99license: mit100---101