An overview of MSC-AD dataset built in a standard imaging system. It covers 12 imaging scenes, a combination of low-to-high three illuminations and four resolutions from 150×150 to 600×600 pixels (Left of red dotted line), and six types of casting surfaces include five kinds of defects (Right of red dotted line)

An overview of open source MSC-AD dataset

This dataset is publicly available. It is free for professors and researcher scientists affiliated to a University.

Permission to use but not reproduce or distribute our database is granted to all researchers given that the following steps are properly followed:
Send an e-mail to mscad2023@gmail.com before downloading the dataset. You will need a password to access the files of this dataset.
Your Email MUST be sent from a valid University account and MUST include the following text:

1.Subject: Application to Download the MSC-AD Dataset
2.Name: [Your First and Last Name]
3.Affiliation: [University where You Work]
4.Department: [Your Department]
5.Position: [Your Job Title]
6.Email: [Email Address from Your Affiliation]
7.Supervisor: [Name of Your Supervisor](necessary if you are a student)
8.Supervisor Email: [Email Address of Your Supervisor](if 7 is provided)

I have read and agree to the terms and conditions specified in the MSC-AD database webpage.
This database will only be used for research purposes.
I will not make any part of this database available to a third party.
I'll not sell any part of this database or make any profit from its use.

    Intelligent detection of product surface defects in the industrial scene is the key to ensuring product quality. On general benchmarks, current unsupervised anomaly detection techniques have achieved significant success. When used in complex industrial environments (e.g., large industrial components with minute defect), the model needs to be able to adapt to different imaging scenarios (e.g., illumination and resolution) and accurately detect and localize anomalies, but its performance is still far from satisfactory. Besides, the complex and unstable optical lighting environment for collecting such data poses major challenges in establishing unified benchmarks for optical lighting and imaging resolution in defect detection. To fill this gap, we build a standard imaging system-based multiscene unsupervised anomaly detection dataset, coined as MSC-AD. In particular, it provides 12 imaging scenes, i.e., a cross combination of low-to-high three illuminations and 150×150 to 600×600 four resolutions, in which six types of large casting surfaces with different structures include five kinds of defects with sample-level and pixellevel precise ground truth. We systematically investigate representative baseline methods and empirical analysis on this dataset to obtain a number of interesting findings, e.g. how to detach from distinctly different imaging scenes, and how to distinguish between subtly normal-anomaly classes. To the best of our knowledge, MSC-AD is the first multi-illumination, multi-resolution, multi-surface and multidefect dataset built in a standard imaging system. The MSCAD dataset and baseline codes will be available.

An overview of MSC-AD construction.

Overview of multi-scene image acquisition system. (a) a prototype of a multi-scene image acquisition system. (b) Schematic diagram of multi-scene image acquisition system. (c) Schematic diagram of the multi-resolution adjustment module. (d) Design schematic diagram of optical lighting module.

Visualization comparisons of different surface defects in 12 scenarios.

Qualitative results for anomaly localization on MSC-AD dataset. (a) Qualitative analysis results for anomaly localization under different resolutions. (b) Qualitative analysis results for anomaly localization under different illuminations. (c) Misjudged results for normal samples. (d) False positive results for anomaly samples.