Can Image Splicing and Copy-Move Forgery Be Detected by the Same Model? Forensim: An Attention-Based State-Space Approach
Soumyaroop Nandi, Prem Natarajan · 2026
Read Can Image Splicing and Copy-Move Forgery Be Detected by the Same Model? Forensim: An Attention-Based State-Space Approach by Soumyaroop Nandi, Prem Natarajan (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-2602-10079.txt.
Description
Image forgery detection usually hunts for artifacts in the altered region alone, but that can mislead, as when a duplicated act is inserted into a peaceful crowd. A 2026 paper by Soumyaroop Nandi and Prem Natarajan introduces Forensim, an attention-based state-space framework that localizes both manipulated target regions and their source regions together. Its masks carry three classes: pristine, source, and target.
A visual state-space model with normalized attention maps finds internal similarities, while a region-based block attention module separates manipulated regions. One architecture handles both splicing and copy-move forgery with end-to-end training. Forensim leads standard benchmarks, and the team releases a new data set, CMFD-Anything, fixing limits of existing copy-move collections.
Catalog record
| CREATOR | Soumyaroop Nandi, Prem Natarajan |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-10079