Detecting Temporally Localized Manipulations in Authentic Video Streams Dataset
Description
This dataset accompanies the paper "Detecting Temporally Localized Manipulations in Authentic Video Streams". It contains 100 authentic source videos, 100 manipulated videos, 100 generated fake video segments, 100 representative frame images, and metadata describing the inserted fake segment interval for each manipulated video.
The dataset is organized by shared example ids. For each example id, the authentic source video, fake segment, manipulated video, representative image, and metadata row correspond to the same sample. The accompanying GitHub repository provides the cleaned evaluation and plotting code used for the reported experiments:
https://github.com/OkanUmur/temporally-localized-video-manipulation-detection
Files
temporally-localized-video-manipulation-detection-dataset.zip
Additional details
Software
- Repository URL
- https://github.com/OkanUmur/temporally-localized-video-manipulation-detection
- Programming language
- Python