From Intuition to Investigation: A Tool-Augmented Reasoning MLLM Framework for Generalizable Face Anti-Spoofing

Haoyuan Zhang, Keyao Wang, Guosheng Zhang, Haixiao Yue, Zhiwen Tan, Siran Peng, Tianshuo Zhang, Xiao Tan, Kunbin Chen, Wei He, Jingdong Wang, Ajian Liu, Xiangyu Zhu, Zhen Lei · 2026

Read From Intuition to Investigation: A Tool-Augmented Reasoning MLLM Framework for Generalizable Face Anti-Spoofing by Haoyuan Zhang, Keyao Wang, Guosheng Zhang, Haixiao Yue, Zhiwen Tan, Siran Peng, Tianshuo Zhang, Xiao Tan, Kunbin Chen, Wei He, Jingdong Wang, Ajian Liu, Xiangyu Zhu, Zhen Lei (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-2603-01038.txt.

Description

Zhang and thirteen colleagues strengthen face anti-spoofing by letting vision-language models investigate, not just glance. Their TAR-FAS framework turns spoof detection into step-by-step reasoning with visual tools. It starts from intuitive cues and calls tools for fine detail when needed. Trained with a tool-aware method on a new sixteen-thousand-trajectory dataset, it leads the field under a harsh one-to-eleven cross-domain test. It also shows its visual evidence for trustworthy decisions.

Catalog record

CREATORHaoyuan Zhang, Keyao Wang, Guosheng Zhang, Haixiao Yue, Zhiwen Tan, Siran Peng, Tianshuo Zhang, Xiao Tan, Kunbin Chen, Wei He, Jingdong Wang, Ajian Liu, Xiangyu Zhu, Zhen Lei
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46871
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.01038

Browse the full library · About the Tower of Babel