Add support for Photometric stereo (PS)-AD dataset #649
Unanswered
jpcbertoldo
asked this question in
Feature Requests
Replies: 1 comment
-
Thanks @jpcbertoldo, looking promising! Moving to feature requests for now. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
[1] recently published a dataset similar to MVTec-AD with photometric stereo images -- each object has 4 images with exactly the same position but with different illumination angles (example below).
Describe the solution you'd like
Create a dataset and datamodule for this dataset.
This should probably wait for #572.
Additional context
The paper (version from the 18 October 2022, link below) says (footnote 1) "Dataset and codes will be available when this paper becomes public.", so I believe it has not yet been published.[update]
The paper's code is available here: github.com/Junyong-Jung/PSAD
The dataset is availbla for download here (~100Gb): cnu365-my.sharepoint.com/:u:/g/personal/jjy9385_o_cnu_ac_kr/EeAfcf2zU9NCoz8KjhpeJA8BA5Xv6gp0yxryw-o5AsSUXw
Size of the train/test sets for the 10 categories in the dataset (numbers in terms of instances of object, the number of images is multiplied by 4).
Source: [1].
Link to the paper on Semantic Scholar: www.semanticscholar.org/paper/Anomaly-Detection-Requires-Better-Representations-Reiss-Cohen/f785d513eb6cef58de1e761534a1598515118faa
Beta Was this translation helpful? Give feedback.
All reactions