Summary of Omni-iml: Towards Unified Image Manipulation Localization, by Chenfan Qu et al.
Omni-IML: Towards Unified Image Manipulation Localization
by Chenfan Qu, Yiwu Zhong, Fengjun Guo, Lianwen Jin
First submitted to arxiv on: 22 Nov 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes Omni-IML, the first generalist model to unify diverse Image Manipulation Localization (IML) tasks. Existing IML methods rely heavily on task-specific designs and perform well only on one target image type but are mostly random guessing on other image types. The proposed Omni-IML achieves generalism by adopting Modal Gate Encoder and Dynamic Weight Decoder to adaptively determine optimal encoding modality and decoder filters for each sample. Additionally, an Anomaly Enhancement module is introduced to enhance features of tampered regions with box supervision. Experimental results demonstrate state-of-the-art performance on three major IML tasks (natural images, document images, and face images) using a single unified model. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research helps keep visual content secure by detecting when pictures are manipulated or altered. Right now, there are special methods to do this, but they only work well for specific types of images. The new approach, called Omni-IML, can handle many different image types and even improve performance when trying to detect tampered areas. The researchers tested it on three types of images (natural, documents, and faces) and found that it worked better than other methods. This could help make the internet a safer place. |
Keywords
» Artificial intelligence » Decoder » Encoder
