OmniVL-Guard: Towards Unified Vision-Language Forgery Detection and Grounding via Balanced RL

Jinjie Shen, Jing Wu, Yaxiong Wang, Lechao Cheng, Shengeng Tang, Tianrui Hui, Nan Pu, Zhun Zhong · 2026

Read OmniVL-Guard: Towards Unified Vision-Language Forgery Detection and Grounding via Balanced RL by Jinjie Shen, Jing Wu, Yaxiong Wang, Lechao Cheng, Shengeng Tang, Tianrui Hui, Nan Pu, Zhun Zhong (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-10687.txt.

Description

Jinjie Shen and seven co-authors unify forgery finding across text, images, and video in a 2026 paper. Real misinformation interleaves modalities, but current detectors handle one or two at a time. A unified setting adds a subtle trap: the easier veracity sorting task dominates slopes, starving fine-grained grounding during multi-task training.

Their system, OmniVL-Guard, uses even reward-based learning to counter that hardness bias. Self-evolving designs rebalance the dual demands of finding and localization so grounding keeps pace with sorting. The paper targets a useful need: not just flagging forged content, but showing where the forgery sits across every modality it touches.

Catalog record

CREATORJinjie Shen, Jing Wu, Yaxiong Wang, Lechao Cheng, Shengeng Tang, Tianrui Hui, Nan Pu, Zhun Zhong
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE87 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-10687

Browse the full library · About the Tower of Babel