Long-Tailed Anomaly Detection (LTAD) Dataset

<p><strong>Introduction</strong></p> <p>Anomaly detection (AD) aims to identify defective images and localize their defects (if any). Ideally, AD models should be able to: detect defects over many image classes; not rely on hard-coded class names that can be uninformative or inconsistent across datasets; learn without anomaly supervision; and be robust to the long-tailed distributions of real-world applications. To address these challenges, we formulate the problem of long-tailed AD by introducing several datasets with different levels of class imbalance for performance evaluation.</p> <p>To encourage more follow up works on long-tailed AD, we are publicly releasing the dataset split used in our paper (&ldquo;Long-Tailed Anomaly Detection with Learnable Class Names&rdquo; by Chih-Hui Ho, Kuan-Chuan Peng, and Nuno Vasconcelos, CVPR 2024).</p> <p>Files in the unzipped folder:</p> <p>1.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; ./README.md: This Markdown file</p> <p>2.&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; ./dataset_split: Folder contains long-tail splits from three datasets. See below for details.</p> <p><strong>&nbsp;</strong></p> <p><strong>At a Glance</strong></p> <ul> <li>The size of the unzipped dataset is ~16MB</li> <li>Three datasets are used in this project, including [MVTec](https://www.mvtec.com/company/research/datasets/mvtec-ad), [VisA](https://github.com/amazon-science/spot-diff) and [DAGM](https://www.kaggle.com/datasets/mhskjelvareid/dagm-2007-competition-dataset-optical-inspection). Please download the datasets from their original repositories.</li> <li>The dataset split provided in this folder is organized as follows:<br>```<br>dataset_split<br>|---dagm_lt<br>|---mvtec_lt<br>|---visa_lt<br>|-----|-- exp<br>|-----|-----|----- 100<br>|-----|-----|----- |-----test.json<br>|-----|-----|----- |-----train.json<br>|-----|-----|----- 200<br>|-----|-- step<br>|-----|-- ...<br>```</li> <li>Each long-tailed dataset split contains a subfolder ``imbalance_type/imbalance_factor", where imbalance type can be [exponential (exp), step, reverse exponential (exp_reverse), reverse step (step_reverse)]. The definition of imbalance type and imbalance factor can be found in our paper. Each subfolder contains two json files, one for training and the other for testing.</li> <li>Each entry in the json file contains the meta information of an image and is similar to<br>```<br>{"filename": "candle/test/bad/000.JPG", "label": 1, "label_name": "defective", "clsname": "candle", "maskname": "candle/ground_truth/bad/000.png"}<br>```<br>- filename: location of the input image in the dataset<br>- label: indicates whether the input image is normal (labeled as 0) or defective (labeled as 1)<br>- label name: can be "good" or "defective"<br>- clsname: class name of the input image<br>- maskname (optional): location of the binary image that indicates the defect region. This is only available for test.json, because there is no defect image during training.</li> </ul> <p><strong>Citation</strong></p> <p>If you use the LTAD dataset in your research, please cite our contribution:</p> <pre><code>@InProceedings{Ho_2024_CVPR, author = {Ho, Chih-Hui and Peng, Kuan-Chuan and Vasconcelos, Nuno}, title = {Long-Tailed Anomaly Detection with Learnable Class Names}, booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)}, month = {June}, year = {2024} } </code></pre> <p><strong>License</strong></p> <p>The LTAD dataset is released under&nbsp;<a href="https://creativecommons.org/licenses/by-sa/4.0/">CC-BY-SA-4.0 license</a>. For the images in the MVTec, VisA, and DAGM datasets, please refer to their websites for their copyright and license terms.</p> <pre><code>Created by Mitsubishi Electric Research Laboratories (MERL), 2023-2024 SPDX-License-Identifier: CC-BY-SA-4.0</code></pre>