A naukri.com initiative
Arxiv
2M
22
Image Credit: Arxiv
ExDDV: A New Dataset for Explainable Deepfake Detection in Video
- ExDDV is a new dataset and benchmark designed to address the need for explainable deepfake detection in video.
- The dataset contains approximately 5.4K real and deepfake videos, with manual annotations of text descriptions and artifact localization clicks.
- Experiments with vision-language models on ExDDV demonstrate the importance of both text and click supervision in developing robust explainable deepfake detection models.
- The dataset and code to reproduce the results are available at https://github.com/vladhondru25/ExDDV.
Read Full Article
1 Like
For uninterrupted reading, download the app