Datasets:

Modalities:
Text
Image
Formats:
parquet
Languages:
English
ArXiv:
Tags:
image
Libraries:
Datasets
Dask
License:
jespark commited on
Commit
005ea06
·
verified ·
1 Parent(s): 7ae5627

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +2 -0
README.md CHANGED
@@ -31,6 +31,8 @@ The following table shows the performance (AP) difference between the classifier
31
  | Base | 0.995 | 0.996 | 0.947 | 0.985 | 0.998 | 0.984 |
32
  | Small | 0.986 | 0.995 | 0.888 | 0.852 | 0.993 | 0.943 |
33
 
 
 
34
  ## Dataset Summary
35
  - The Community Forensics (small) dataset is intended for developing and benchmarking forensics methods that detect or analyze AI-generated images. It contains 278K generated images collected from 4803 generator models, and paired with 278K "real" images, sourced from [FFHQ](https://github.com/NVlabs/ffhq-dataset), [VISION](https://lesc.dinfo.unifi.it/VISION/), [COCO](https://cocodataset.org/), and [Landscapes HQ](https://github.com/universome/alis) datasets.
36
 
 
31
  | Base | 0.995 | 0.996 | 0.947 | 0.985 | 0.998 | 0.984 |
32
  | Small | 0.986 | 0.995 | 0.888 | 0.852 | 0.993 | 0.943 |
33
 
34
+ *Please check [Community Forensics-Eval](https://huggingface.co/datasets/OwensLab/CommunityForensics-Eval) for the recommended 'comprehensive' evaluation set.*
35
+
36
  ## Dataset Summary
37
  - The Community Forensics (small) dataset is intended for developing and benchmarking forensics methods that detect or analyze AI-generated images. It contains 278K generated images collected from 4803 generator models, and paired with 278K "real" images, sourced from [FFHQ](https://github.com/NVlabs/ffhq-dataset), [VISION](https://lesc.dinfo.unifi.it/VISION/), [COCO](https://cocodataset.org/), and [Landscapes HQ](https://github.com/universome/alis) datasets.
38