Skip to content

Blue-radish/MMFN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MMFN / SECMME 2025

本仓库为论文 《Fake Information Detection Method Based on BERT Pre-trained Model》 的官方代码实现,对应第二十六届中国机器人及人工智能大赛参赛项目。
The code implementation for the article "Fake Information Detection Method Based on BERT Pre-trained Model".

论文链接(IEEE):https://ieeexplore.ieee.org/abstract/document/11138502


📌 项目简介

MMFN是一个面向微博场景的多模态虚假信息检测系统,结合 文本语义特征 + 图像视觉特征,通过跨模态对齐机制实现对图文不一致、伪造内容的识别。


📁 仓库结构

MMFN
│  .gitignore
│  main.ipynb
│  main.py
│  README.md
│  requirements.txt
│
├─figure
│      bert.jpg
│      bert_1.png
│      WebUI.png
│
├─flagged
│      log.csv
│
├─models
│  ├─hfl_rbt6
│  │
│  └─img_model
│          resnet50.pth
│
├─notebook
│      data.ipynb
│      train.ipynb
│      train_all.ipynb
│      train_text.ipynb
│
├─src
│      model.py
│      train.py
│
└─temp                   # 临时文件(如爬取的微博图文)

🚀 快速开始

安装依赖

pip install -r requirements.txt

建议使用 Python 3.10 与 GPU 环境(CUDA)。

数据预处理

data.ipynb

训练

  • train.ipynb (multi-modal, can handle multiple images)
  • train_all.ipynb (multi-modal, if there are multiple images, only one will be used)
  • train_text.ipynb (uni-modal, using only text information)

启动WebUI

python main.py

🧠 模型结构

Text Encoder (hfl_rbt6)        Image Encoder (ResNet-50)
          ↓                               ↓
     Text Embedding                 Visual Embedding
          ↘                             ↙
        Multi-head Cross-modal Attention
                    ↓
               Fusion Vector
                    ↓
                Classifier

Poster (多模态的虚假信息检测系统)

Poster.

About

A multimodal fake news detection system using BERT and ResNet50. Includes a web UI for easy inference.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors