A new 140K-video dataset and native-scale Qwen2.5-VL detector for AI-generated videos that avoids preprocessing losses to retain high-frequency forgery artifacts.
The models differ significantly in training methodology, data scale, output resolution, and video duration, contributing to a richly diverse training set
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Preserving Forgery Artifacts: AI-Generated Video Detection at Native Scale
A new 140K-video dataset and native-scale Qwen2.5-VL detector for AI-generated videos that avoids preprocessing losses to retain high-frequency forgery artifacts.