ComfyUI_DeepFakeDefenders

ComfyUI_DeepFakeDefenders
★ 43

深度伪造检测ComfyUI节点阈值判定图像裁切
在 ComfyUI 中集成 DeepFakeDefenders 模型，判断图像是否为 DeepFake；支持阈值判定、512×512裁切以提升识别，未匹配时输出带提示的空图像。
💡 在 ComfyUI 流程中自动识别并标记可能的 DeepFake 图像。
🍴 2 Forks💻 Python🔄 2024-09-14
📦
网盘下载
复制链接后前往夸克网盘下载
https://pan.quark.cn/s/c1eafc754fbb
📦 requirements.txt
#torch
#torchvision
#transformers
#cv2
📄 README

ComfyUI_DeepFakeDefenders

You can using DeepFakeDefenders in comfyUI to Prediction image is a DeepFake img or not.

from 1st place solution for The Global Multimedia Deepfake Detection (Image Track) by “JTGroup” team.

DeepFakeDefenders From: link

Update

2024/09/13

  • Add the function of distinguishing images based on thresholds/增加根据阈值区分图片的功能
  • Unmatched image will get a empty output/如果没有匹配的图片，会得到一张带有文字提示的空输出。
  • Add image cropping function to better match the model/加入图片裁切，方法内置是512*512的分块，识别精度有轻微提升

    • 1.Installation

    In the ./ComfyUI /custom_node directory, run the following:

    git clone https://github.com/smthemex/ComfyUI_DeepFakeDefenders.git

    2.requirements

    For ComfyUI users, all libraries in the requirements file should be available. If not, please uncomment the # and reinstall

    pip install -r requirements.txt

    3 Need model

    download models 百度云 or google drive

    ├── ComfyUI/models/DeepFakeDefender
|             ├── ema.state
|             ├── weight.pth

    4 Example

    Notice,example img are all deepfake img,so we still need to improve recognition accuracy

    5 Citation

    This work is licensed under a link License

    DeepFakeDefenders From: link