This is the inverted mask dataset of the testing set of NVIDIA Irregular Mask Dataset.
- 12,000 images: 12 categories of 1,000 images each.
- Image size: 512 x 512 (RGB)
- File format: JPG
The images are grouped in 12 categories with 6 hole-to-image area ratios (HTIAR), with and without border.
File name | HTIAR | Note |
---|---|---|
00000.jpg - 00999.jpg | 0% - 10% | with border |
01000.jpg - 01999.jpg | 0% - 10% | no border |
02000.jpg - 02999.jpg | 10% - 20% | with border |
03000.jpg - 03999.jpg | 10% - 20% | no border |
04000.jpg - 04999.jpg | 20% - 30% | with border |
05000.jpg - 05999.jpg | 20% - 30% | no border |
06000.jpg - 06999.jpg | 30% - 40% | with border |
07000.jpg - 07999.jpg | 30% - 40% | no border |
08000.jpg - 08999.jpg | 40% - 50% | with border |
09000.jpg - 09999.jpg | 40% - 50% | no border |
10000.jpg - 10999.jpg | 50% - 60% | with border |
11000.jpg - 11999.jpg | 50% - 60% | no border |
Zip file in Google Drive.
"This dataset is for research purposed only."
NVIDIA Irregular Mask Dataset: https://nv-adlr.github.io/publication/partialconv-inpainting