PROMPT-IML: Image Manipulation Localization with Pre-trained Foundation Models Through Prompt Tuning
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911744824180736 |
|---|---|
| author | Liu, Xuntao Yang, Yuzhou Ying, Qichao Qian, Zhenxing Zhang, Xinpeng Li, Sheng |
| author_facet | Liu, Xuntao Yang, Yuzhou Ying, Qichao Qian, Zhenxing Zhang, Xinpeng Li, Sheng |
| contents | Deceptive images can be shared in seconds with social networking services, posing substantial risks. Tampering traces, such as boundary artifacts and high-frequency information, have been significantly emphasized by massive networks in the Image Manipulation Localization (IML) field. However, they are prone to image post-processing operations, which limit the generalization and robustness of existing methods. We present a novel Prompt-IML framework. We observe that humans tend to discern the authenticity of an image based on both semantic and high-frequency information, inspired by which, the proposed framework leverages rich semantic knowledge from pre-trained visual foundation models to assist IML. We are the first to design a framework that utilizes visual foundation models specially for the IML task. Moreover, we design a Feature Alignment and Fusion module to align and fuse features of semantic features with high-frequency features, which aims at locating tampered regions from multiple perspectives. Experimental results demonstrate that our model can achieve better performance on eight typical fake image datasets and outstanding robustness. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_00653 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | PROMPT-IML: Image Manipulation Localization with Pre-trained Foundation Models Through Prompt Tuning Liu, Xuntao Yang, Yuzhou Ying, Qichao Qian, Zhenxing Zhang, Xinpeng Li, Sheng Computer Vision and Pattern Recognition Deceptive images can be shared in seconds with social networking services, posing substantial risks. Tampering traces, such as boundary artifacts and high-frequency information, have been significantly emphasized by massive networks in the Image Manipulation Localization (IML) field. However, they are prone to image post-processing operations, which limit the generalization and robustness of existing methods. We present a novel Prompt-IML framework. We observe that humans tend to discern the authenticity of an image based on both semantic and high-frequency information, inspired by which, the proposed framework leverages rich semantic knowledge from pre-trained visual foundation models to assist IML. We are the first to design a framework that utilizes visual foundation models specially for the IML task. Moreover, we design a Feature Alignment and Fusion module to align and fuse features of semantic features with high-frequency features, which aims at locating tampered regions from multiple perspectives. Experimental results demonstrate that our model can achieve better performance on eight typical fake image datasets and outstanding robustness. |
| title | PROMPT-IML: Image Manipulation Localization with Pre-trained Foundation Models Through Prompt Tuning |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2401.00653 |