ForensicZip: More Tokens are Better but Not Necessary in Forensic Vision-Language Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912963534782464 |
|---|---|
| author | Lai, Yingxin Yu, Zitong Wang, Jun Shen, Linlin Xu, Yong Cao, Xiaochun |
| author_facet | Lai, Yingxin Yu, Zitong Wang, Jun Shen, Linlin Xu, Yong Cao, Xiaochun |
| contents | Multimodal Large Language Models (MLLMs) enable interpretable multimedia forensics by generating textual rationales for forgery detection. However, processing dense visual sequences incurs high computational costs, particularly for high-resolution images and videos. Visual token pruning is a practical acceleration strategy, yet existing methods are largely semantic-driven, retaining salient objects while discarding background regions where manipulation traces such as high-frequency anomalies and temporal jitters often reside. To address this issue, we introduce ForensicZip, a training-free framework that reformulates token compression from a forgery-driven perspective. ForensicZip models temporal token evolution as a Birth-Death Optimal Transport problem with a slack dummy node, quantifying physical discontinuities indicating transient generative artifacts. The forensic scoring further integrates transport-based novelty with high-frequency priors to separate forensic evidence from semantic content under large-ratio compression. Experiments on deepfake and AIGC benchmarks show that at 10\% token retention, ForensicZip achieves $2.97\times$ speedup and over 90\% FLOPs reduction while maintaining state-of-the-art detection performance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_12208 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ForensicZip: More Tokens are Better but Not Necessary in Forensic Vision-Language Models Lai, Yingxin Yu, Zitong Wang, Jun Shen, Linlin Xu, Yong Cao, Xiaochun Computer Vision and Pattern Recognition Multimodal Large Language Models (MLLMs) enable interpretable multimedia forensics by generating textual rationales for forgery detection. However, processing dense visual sequences incurs high computational costs, particularly for high-resolution images and videos. Visual token pruning is a practical acceleration strategy, yet existing methods are largely semantic-driven, retaining salient objects while discarding background regions where manipulation traces such as high-frequency anomalies and temporal jitters often reside. To address this issue, we introduce ForensicZip, a training-free framework that reformulates token compression from a forgery-driven perspective. ForensicZip models temporal token evolution as a Birth-Death Optimal Transport problem with a slack dummy node, quantifying physical discontinuities indicating transient generative artifacts. The forensic scoring further integrates transport-based novelty with high-frequency priors to separate forensic evidence from semantic content under large-ratio compression. Experiments on deepfake and AIGC benchmarks show that at 10\% token retention, ForensicZip achieves $2.97\times$ speedup and over 90\% FLOPs reduction while maintaining state-of-the-art detection performance. |
| title | ForensicZip: More Tokens are Better but Not Necessary in Forensic Vision-Language Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.12208 |