LEGION

KO | EN

gitlite — search

#typescript #ai-agents #deepseek-harness #dsh-plugin #open-source #ai #claude-code #cli #windows #codex #developer-tools #dsh

★ 82 Open GitHub ↗

[ICCV25 Highlight] The official implementation of the paper "LEGION: Learning to Ground and Explain for Synthetic Image Detection"

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<div align="center">
<h2> LEGION: Learning to Ground and Explain for Synthetic Image Detection
</h2> 
<div align="center">

[Hengrui Kang](https://scholar.google.com/citations?user=kVbzWCAAAAAJ&hl=zh-CN)\*, [Siwei Wen](https://scholar.google.com/citations?user=kJRiUYwAAAAJ&hl=zh-CN)\*, [Zichen Wen](https://scholar.google.com/citations?user=N-aPFvEAAAAJ&hl=zh-CN)\*, [Junyan Ye](https://yejy53.github.io/), [Weijia Li](https://liweijia.github.io/)<sup>†</sup><br>[Peilin Feng](https://peilin-ff.github.io/), [Baichuan Zhou](https://baichuanzhou.github.io/), [Bin Wang](https://wangbindl.github.io/), [Dahua Lin](http://dahua.site/), [Linfeng Zhang](http://www.zhanglinfeng.tech/), [Conghui He](https://conghui.github.io/)<sup>†</sup>

</div>
<a href="https://arxiv.org/pdf/2503.15264" target="_blank"><img src="https://img.shields.io/badge/arXiv-Paper-red?style=badge&logo=arXiv" alt="Paper PDF" height="25"></a>&nbsp;&nbsp;&nbsp;&nbsp;
<a href="https://opendatalab.github.io/LEGION/" target="_blank"><img alt="Website" src="https://img.shields.io/badge/🌎_Project Page-Website-blue.svg" height="25" /></a>&nbsp;&nbsp;&nbsp;&nbsp;
<a href='https://huggingface.co/datasets/khr0516/SynthScars'><img src='https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Dataset-yellow' height="25"></a>
</div>

## 📢 Latest News
- **[2025.10.22]**: 🔥 Due to storage changes, unfortunately I can’t locate the original final weights. However, I'm able to find intermediate [weights](https://huggingface.co/khr0516/legion_LE) that can perform artifact localization and explanation. Please do it yourself for the detection task.
- **[2025.07.24]**: 🔥 **LEGION has been accepted as a "Highlight" paper at ICCV 2025!** 🎉🎉🎉
- **[2025.03.21]**: 🔥 Now, **SynthScars dataset is available** on HuggingFace. Click [here](https://huggingface.co/datasets/khr0516/SynthScars) to download it.
- **[2025.03.20]**: We have released our paper on arXiv. Check out the paper [here](https://arxiv.org/pdf/2503.15264).

## <img id="painting_icon" width="3%" src="assets/icons/robot.png"> &nbsp;Overview

The rapid advancements in generative technology have emerged as a double-edged sword. While offering powerful tools that enhance convenience, they also pose significant social concerns. As **defenders**, current synthetic image detection methods often lack artifact-level textual interpretability and are overly focused on image manipulation detection, and current datasets usually suffer from outdated generators and a lack of fine-grained annotations. In this paper, we introduce **SynthScars**, a high-quality and diverse dataset consisting of 12,236 fully synthetic images with human-expert annotations. It features 4 distinct image content types, 3 categories of artifacts, and fine-grained annotations covering pixel-level segmentation, detailed textual explanations, and artifact category labels. Furthermore, we propose **LEGION** (**LE**arning to **G**round and explain for Synthetic **I**mage detecti**ON**), a multimodal large language model (MLLM)-based image forgery analysis framework that integrates artifact detection, segmentation, and explanation. Building upon this capability, we further explore LEGION as a controller, integrating it into image refinement pipelines to guide the generation of higher-quality and more realistic images. Extensive experiments show that LEGION outperforms existing methods across multiple benchmarks, particularly surpassing the second-best traditional expert on SynthScars by **3.31%** in mIoU and **7.75%** in F1 score. Moreover, the refined images generated under its guidance exhibit stronger alignment with human preferences. The code, model, and dataset will be released.

## 🏆 Contribution

- We introduce **SynthScars**, a **challenging dataset** for synthetic image detection, featuring high-quality synthetic images with diverse content types, as well as fine-grained pixel-level artifact annotations with detailed textual explanations.
- We propose **LEGION**, a comprehensive **image forgery analysis framework** for artifact localization, explanation generation, and forgery detection, which effectively aids human experts to detect and understand image forgeries.
- Extensive experiments demonstrate that LEGION achieves **exceptional performance** on **4** challenging benchmarks. Comparisons with **19** existing methods show that it achieves SOTA performance on majority of metrics, exhibiting strong **robustness** and **generalization** ability.
- We position LEGION not only as a **defender** against ever-evolving generative technologies but also as a **controller** that guides **higher-quality and more realistic image generation**.

## 🛠️ Installation
Please clone our repository and change directory to the folder:
```bash
git clone git@github.com:opendatalab/LEGION.git
cd LEGION
```

Our model is based on the [GLaMM](https://github.com/mbzuai-oryx/groundingLMM/blob/main/docs/install.md) environment. Please follow the steps below to configure the environment.
```bash
conda create -n legion python=3.10
conda activate legion
pip install -r requirements/default.txt

# Install mmcv
git clone https://github.com/open-mmlab/mmcv
cd mmcv
git checkout v1.4.7
MMCV_WITH_OPS=1 pip install -e .
```

## 📦 Dataset

**SynthScars** is a high-quality and diverse dataset consisting of **12,236** fully synthetic images with human-expert annotations. It features **4** distinct image content types, **3** categories of artifacts, and fine-grained annotations covering **pixel-level segmentation**, **detailed textual explanations**, and **artifact category labels**.  Please download it from [HuggingFace](https://huggingface.co/datasets/khr0516/SynthScars), unzip it into the `./data` folder, and ensure it follows the structure below:
```
./data
└── SynthScars
    ├── train
    │   ├── images
    │   │   ├── xxx.jpg
    │   │   └── ...
    │   └── annoations
    │       └── train.json
    └── test
        ├── images
        │   ├── xxx.jpg
        │   └── ...
        └── annoations
            └── test.json
```

## 📌 Usage
As a **defender**, we provide the training and evaluation code for the image forgery analysis task as follows:

- ### <img id="painting_icon" width="2.2%" src="assets/icons/localization.png"> Localization and &nbsp;<img id="painting_icon" width="2.2%" src="assets/icons/explanation.png"> Explanation
  If you want to train LEGION to perform artifact localiztion and explanation generation, please first download the pretrained weights of [GLaMM](https://huggingface.co/MBZUAI/GLaMM-GranD-Pretrained) & [SAM](https://huggingface.co/spaces/abhishek/StableSAM/blob/main/sam_vit_h_4b8939.pth), and then run:
  ```bash
    bash scripts/loc_exp/train.sh
  ```
  After training, you need to merge the weights as follows:
  ```bash
    bash scripts/merge_weights/step1.sh
    bash scripts/merge_weights/step2.sh
  ```
  If you want to infer on your data, please put them into one folder and then run:
  ```bash
    bash scripts/loc_exp/infer.sh
  ```
- ### <img id="painting_icon" width="2.2%" src="assets/icons/detection.png"> Detection
  If you want to train LEGION to conduct deepfake detection, **please replace the data path and model path** in the bash file and run:
  ```bash
    bash scripts/cls/train.sh
  ```
  If you want to eval on your data, please run:
  ```bash
    bash scripts/cls/eval.sh
  ```

As a **controller**, we provide the pipelines of both image regeneration and inpainting for high-quality and more realistic image generation:

- ### Regeneration
  You can run iteratively prompt revision and image regeneration via:
  ```bash
    conda create -n regenerator python=3.10 -y
    conda activate regenerator
    pip install -r requirements/regenerator.txt
    bash scripts/refine/regenerator.sh
  ```
- ### Inpainting
  You can run iteratively regional inpainting via:
  ```bash
    bash scripts/refine/inpainter.sh
  ```

## 📊 Performance
As a **defender**, we evaluate LEGION's performance on three key subtasks: **artifact localization**, **explanation generation**, and **deepfake detection**. 
- ### <img id="painting_icon" width="2.2%" src="assets/icons/localization.png"> Localization
    We evaluate the artifact localization task using the **SynthScars**, **RichHF-18K**, and **LOKI** datasets, which contain <u>region-level mask</u> annotations.
<div align="center">
<img src="assets/tables/localization_table.png" alt="result" width="90%" height="auto">
</div>

- ### <img id="painting_icon" width="2.2%" src="assets/icons/explanation.png"> Explanation 
    To assess the model's natural language interpretability, we use the **SynthScars** and **LOKI** datasets, which provide <u>detailed artifact explanations</u>.
<div align="center">
<img src="assets/tables/explanation_table.png" alt="result" width="80%" height="auto">
</div>

- ### <img id="painting_icon" width="2.2%" src="assets/icons/detection.png"> Detection
    For the deepfake detection task, we adopt the classic **UniversialFakeDetect** benchmark.
<div align="center">
<img src="assets/tables/detection_table.png" alt="result" width="80%" height="auto">
</div>

As a **controller**, we explore LEGION's capability to guide image generation through two pipelines: image **regeneration** and **inpainting**, based on its analytical feedback.
<div align="center">
<img src="assets/tables/hps_table.png" alt="result" width="60%" height="auto">
</div>

**Note**: More visualization results and case studies please refer to our [Project Page](https://opendatalab.github.io/LEGION/).

## 😄 Acknowledgement

This repository is built upon the work of [GLaMM](https://github.com/mbzuai-oryx/groundingLMM). We appreciate their contributions and insights that have provided a strong foundation for our research.

## 📜 Citation
If you find our work interesting and helpful, please consider giving our repo a star. Additionally, if you would like to cite our work, please use the following format:
```bibtex
@article{kang2025legion,
  title={Legion: Learning to ground and explain for synthetic image detection},
  author={Kang, Hengrui and Wen, Siwei and Wen, Zichen and Ye, Junyan and Li, Weijia and Feng, Peilin and Zhou, Baichuan and Wang, Bin and Lin, Dahua and Zhang, Linfeng and others},
  journal={arXiv preprint arXiv:2503.15264},
  year={2025}
}
```

LEGION

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README