Deepfakes Detection

🇰🇷 한국어 버전 | 🇯🇵 日本語版 | 📈 Model Evaluation | 🤗 Try Demo | 🤗 Hugging Face

📌 Contents

• 🗂 Project Structure - Directory layout of the package, web app, and docs
• 🐳 Docker Quick Start - Run the full stack (MySQL, Redis, FastAPI, Celery, React) with Docker Compose
• 🤗 Try It Live: Hugging Face Spaces - No-install browser demos for image, video, and XAI deepfake detection
• 📚 DeepFake Video BenchMark Datasets — Overview of Celeb-DF-v2, FF++, and KoDF datasets used for training.
• ⚙️ Data Preparation — Efficient face detection and landmark extraction pipeline using YOLOv8
• 🏗 Model Architecture — Detailed look into our hybrid CNN-ViT (MS-EffViT & MS-EffGCViT) designs.
• 🧬 Model Zoo — Comparison of model variants, parameter counts, and computational complexity (FLOPs).
• 🕐 Model Inference Latency - Fast (b0) vs Pro (b5) inference speed on a single Tesla T4
• 🚀 Training - Step-by-step training scrips with Goolge Colab and W&B experiment tracking
• 📈 Model Evaluation - Benchmarking results
• 💻 Model Usage - Load pretrained models via pip install deepguard or straight from the Hugging Face Hub
• 🔮 Predict Image & Video - Simple Inference examples for detecting deepfakes in image and video
• 🎨 DeepFake AI Explainability - Visualizing model focus using Grad-CAM and attention maps
• 📓 Tutorials - Hands-on Colab notebooks for inference and dual-branch XAI visualization
• 📬 Authors - Team behind this senior graduation project at Chungbuk National University
• 📝 Reference - Libraries, datasets, and prior work this project builds on
• ⚖️ License - MIT license

🗂 Project Structure

DeepGuard/
├── deepguard/                  # Core package (models, layers, data pipeline, optimizer, scheduler, utils)
├── preprocess/                 # Face detection, cropping & landmark extraction pipeline
├── inference/                  # Image / video prediction
├── explainability/             # Dual-branch CAM-based XAI
├── App/                        # FastAPI backend (excluded from PyPI package)
├── client/                     # React frontend
├── docker/                     # MySQL / Redis Docker images and configs
├── tutorials/                  # Colab notebooks (prediction & XAI visualization)
├── docs/                       # README assets (architectures, ablation, benchmarks, samples, XAI results)
├── .github/                    # CI workflows and repository settings
├── docker-compose.yml          # Full stack (MySQL, Redis, FastAPI, Celery, React)
├── train_eff_vit.py            # Training entrypoint for MS-EffViT
├── train_eff_gcvit.py          # Training entrypoint for MS-EffGCViT
├── pyproject.toml              # Package build config
├── requirements.txt            # Dev environment dependencies
├── README.md, README_KR.md, README_JP.md
└── LICENSE

🐳 Docker Quick Start

Run the full stack (MySQL + Redis + FastAPI + Celery + React) with Docker Compose — no local Python/Node setup required.

Prerequisites: Docker with Compose v2

git clone https://github.com/HanMoonSub/DeepGuard.git
cd DeepGuard
docker compose up -d

Once all containers are up, open http://localhost:80 in your browser. Images are also mirrored to GitHub Packages.

🤗 Try It Live: Hugging Face Spaces

No install, no GPU, no docker compose up — just click and try DeepGuard straight from your browser.

🖼️ Image Detection Upload an image → real / fake probability

🎬 Video Detection Upload a video → frame-aggregated probability

🎨 Detection XAI See why — dual-branch Grad-CAM heatmaps

_{💛 Enjoying the demos? Please leave a ❤️ like on the Space — it means a lot to us!
Looking for the underlying checkpoints instead of the demo UI? Jump to 🤗 Model Usage → Hugging Face Hub.}

📚 DeepFake Video BenchMark Datasets

To evaluate the generalization and robustness of our deepfake detection model, we utilize three large-scale, widely recognized benchmark datasets. Each dataset presents unique challenges and covers different types of forgery methods.

⚙️ Data Preparation

Our preprocessing pipeline is designed to efficiently extract facial features from videos and prepare them for high-accuracy deepfake detection.

Detect Original Face

To maximize preprocessing efficiency, face detection is performed only on original (real) videos. Since mnipulated videos in DeepFake Video BenchMark Datasets share the same spatial coordinates as their sources, these bounding boxes are reused for the corresponding deepfake versions.

🚀 Efficiency Optimizations

• Lightweight Model: Uses yolov8n-face for high-speed inference without sacrificing accuracy.

• Targeted Processing: By detecting faces only in original videos, the total detection workload is reduced by approximately 80%.

• Dynamic Rescaling: To maintain consistent inference speed across different resolutions, frames are automatically resized based on their dimensions:

Frame Size(Longest Side)	Scale Factor	Action
< 300px	2.0
300px - 700px	1.0
700px - 1500px	0.5
> 1500px	0.33

Face Cropping & Landmark Extraction

This module extracts face crops from both original and deepfake videos using the bounding boxes generated in the previous step. It also performs landmark detection to facilitate advanced augmentations like Landmark-based Cutout

🛠 Key Features

• Dynamic Margin with Jitter: Adds a configurable margin around the face. The margin_jitter parameter introduces random variance to the crop size, making the model more robust to different face scales.

• Landmark Localization: Detects 5 primary facial landmarks (eyes, nose, mouth corners) and saves them as .npy files.

DATA_ROOT/
├── crops/
│   └── {video_id}/
│       ├── 12.png
│       └── ...
├── landmarks/
│   └── {video_id}/
│       ├── 12.npy
│       └── ...
└── train_frame_metadata.csv

Dataset-Specific Pipelines

Click the links below to view the specific preprocessing details for each dataset:

• Celeb-DF V2 Preprocess
• FaceForensics++ Preprocess
• KoDF Preprocess

🏗 Model Architecture

Multi Scale Efficient Global Context Vision Transformer is an optimized multi-scale hybrid architecture that integrates CNN-driven spatial inductive bias with hierarchical attention mechanisms to effectively identify subtle(local) artifacts and macro(global) artifacts for robust deepfake forensics."

Explore More Details

• Model Architecture: MS-EffViT - Multi Scale Efficient Vision Transformer

• Advanced Architecture: MS-EFFGCViT - Multi Scale Efficient Global Context Vision Transformer

We utilizes two distinct types of self-attention to capture both long-range and short-range information across feature maps.

• Local Window Attention: this model efficiently captures local textures and precise spatial details while maintaining linear computational complexity relative to the image size.

• Global Window Attention: Unlike Swin Transformer, this module utilizes global-queries that interact with local window keys and values. This allows each local region to incorporate global context, effectively capturing long-range dependencies and providing a comprehensive understanding of the entire spatial structure

🧬 Model Zoo

🕐 Model Inference Latency

Latency is measured with batch size 1 on a single Tesla T4 and video (num_frames=20).

🚀 Training

These training scripts run against the full repo, not the pip install deepguard package. Clone it and install the full dev environment first:

git clone https://github.com/HanMoonSub/DeepGuard.git
cd DeepGuard
pip install -r requirements.txt

We provide training scripts for both ms_eff_vit and ms_eff_gcvit. We recommend using Google Colab for free GPU access and Weightes & Biases(W&B) for experiment tracking

!python -m train_eff_vit \ # train_eff_gcvit
    --root-dir DATA_ROOT \ 
    --model-ver "ms_eff_vit_b5" \ # ms_eff_vit_b0, ms_eff_vit_b5, ms_eff_gcvit_b0, ms_eff_gcvit_b5
    --dataset "ff++" \ # ff++, celeb_df_v2, kodf
    --seed 2025 \ # for reproducibility
    --wandb-api-key "your-api-key" # Write your own api key

📈 Model Evaluation

!python -m inference.predict_video \
    --root-dir DATA_ROOT \
    --margin-ratio 0.2 \
    --conf-thres 0.5 \
    --min-face-ratio 0.01 \
    --model-name "ms_eff_gcvit_b0" \ # ms_eff_vit_b0, ms_eff_vit_b5, ms_eff_gcvit_b0, ms_eff_gcvit_b5
    --model-dataset "kodf" \ # ff++, celeb_df_v2, kodf
    --num-frames 20 \
    --tta-hflip 0.0 \
    --agg-mode "conf" \

Celeb DF(v2) Pretrained Models

FaceForensics++ Pretrained Models

KoDF Pretrained Models

💻 Model Usage

Both options load the raw model only — no face detection/cropping is applied. For end-to-end inference on a real image/video file, see 🔮 Predict Image & Video below.

Available Datasets: celeb_df_v2, ff++, kodf

📦 Via pip (deepguard / timm)

pip install deepguard

Direct Import (via DeepGuard)

from deepguard import ms_eff_gcvit_b0, ms_eff_gcvit_b5

model = ms_eff_gcvit_b0(pretrained=True, dataset="celeb_df_v2")
model = ms_eff_gcvit_b5(pretrained=True, dataset="ff++")

Using timm Interface

import timm
import deepguard

model = timm.create_model("ms_eff_gcvit_b0", pretrained=True, dataset="ff++")
model = timm.create_model("ms_eff_gcvit_b5", pretrained=True, dataset="kodf")

🤗 Via Hugging Face Hub

Every checkpoint is also mirrored to the Hugging Face Hub under KoreaPeter as its own transformers-compatible repo (config + custom modeling code + safetensors weights) — usable directly via the transformers pipeline API with trust_remote_code=True, no deepguard install required.

💛 Find a checkpoint useful? Please leave a ❤️ like on its model card — it means a lot to us!

Model	Celeb-DF-v2	FaceForensics++	KoDF
⚡ ms_eff_gcvit_b0
🔥 ms_eff_gcvit_b5

from transformers import pipeline

# 🖼️ Image classification
clf = pipeline(
    "image-classification",
    model="KoreaPeter/ms-eff-gcvit-deepfake-b0-kodf",  # swap for any model card above
    trust_remote_code=True,
)
result = clf("face.jpg")
# [{'label': 'fake', 'score': 0.9712}, {'label': 'real', 'score': 0.0288}]

# 🎬 Video classification
clf = pipeline(
    "video-classification",
    model="KoreaPeter/ms-eff-gcvit-deepfake-b0-kodf",
    trust_remote_code=True,
)
result = clf("video.mp4", num_frames=20, agg_mode="conf")
# [{'label': 'fake', 'score': 0.9634}, {'label': 'real', 'score': 0.0366}]

🔮 Predict Image & Video

Predict DeepFake Image

from inference.image_predictor import ImagePredictor

# Initialize the predictor
predictor = ImagePredictor(
            margin_ratio = 0.2, #  Margin ratio around the detected face crop
            conf_thres = 0.5, # Confidence threshold for face detection
            min_face_ratio = 0.01, # Minimum face-toframe size ratio to process 
            model_name = "ms_eff_vit_b0", #  ms_eff_vit_b5, ms_eff_gcvit_b0, ms_eff_gcvit_b5  
            dataset = "celeb_df_v2" # ff++, kodf
            )

# Run Inference
result = predictor.predict_img(
            img_path="path/to/image.jpg",
            tta_hflip=0.0 # Horizontal Flip for Test-Time Augmentation 
            )

print(f"Deepfake Probability: {result:.4f}")

Predict DeepFake Video

from inference.video_predictor import VideoPredictor

# Initialize the predictor
predictor = VideoPredictor(
            margin_ratio = 0.2, #  Margin ratio around the detected face crop
            conf_thres = 0.5, # Confidence threshold for face detection
            min_face_ratio = 0.01, # Minimum face-toframe size ratio to process 
            model_name = "ms_eff_vit_b0", #  ms_eff_vit_b5, ms_eff_gcvit_b0, ms_eff_gcvit_b5  
            dataset = "celeb_df_v2" # ff++, kodf
            )

# Run Inference
result = predictor.predict_video(
            video_path = "path/to/video.mp4",
            num_frames = 20, # Number of frames to sample per video
            agg_mode = "conf", # Aggregation Method: 'conf', 'mean', 'vote'
            tta_hflip=0.0 # Horizontal Flip for Test-Time Augmentation 
            )

print(f"Deepfake Probability: {result:.4f}")

🎨 DeepFake AI Explainability

Deepfake detection is only as trustworthy as its explanations. DeepGuard integrates a production-ready XAI Toolkit that visualizes where and why the model flags a face as manipulated — turning a black-box score into actionable forensic evidence.

⭐ Validated on hybrid CNN-ViT architectures, specifically MS-EffViT and MS-EffGCViT.
⭐ Dual-Branch Analysis: Dual-branch design mirrors the model's own multi-scale reasoning

🧠 How Dual-Branch XAI Works

Branch	Feature Map	Focus	Best For
	High Resolution	Local Forgery artifacts	Skin texture, boundary blending, compression traces
	Low Resolution	Global Semantic Structure	Lighting inconsistency, facial geometry, Shadow artifacts

📐 XAI Methods

Each method is assigned to the branch where it performs best empirically.

• aug_smooth applies TTA (horizontal flips) before averaging CAMs → smoother, more object-aligned maps
• eigen_smooth applies PCA noise reduction → retains dominant forgery pattern only

💡 DeepFake XAI Usage

pip install deepguard

Low-Level Branch — Local Artifact Detection

from explainability import HiResCAMExplainer, GradCAMElementWiseExplainer, LayerCAMExplainer

explainer = HiResCAMExplainer(
    model_name   = "ms_eff_gcvit_b0",  # or ms_eff_vit_b0, ms_eff_gcvit_b5, ms_eff_vit_b5
    dataset      = "celeb_df_v2",       # or ff++, kodf
    branch_level = "low",
)

High-Level Branch — Global Semantic Detection

from explainability import EigenGradCAMExplainer, GradCAMPlusPlusExplainer, XGradCAMExplainer

explainer = EigenGradCAMExplainer(
    model_name   = "ms_eff_gcvit_b0",
    dataset      = "celeb_df_v2",
    branch_level = "high",
)

🎨 Visualization Modes

1. Heatmap — Continuous activation distribution

result = explainer.display_heatmap_on_image(
    img_path     = "path/to/image.jpg",
    category     = 1,      # 0: Real, 1: Fake
    threshold    = 0.5,    # binarization cutoff (0.5~1.0), or "auto" for Otsu
    image_weight = 0.5,    # 0.0: heatmap only ← → 1.0: original only
    aug_smooth   = False,  # TTA smoothing (not supported on 'pro' models)
    eigen_smooth = False,  # PCA noise reduction
)

2. Bounding Box — Discrete forgery region localization

result = explainer.display_bbox_on_image(
    img_path     = "path/to/image.jpg",
    category     = 1,
    threshold    = 0.5,
    thickness    = 1,
    aug_smooth   = False,
    eigen_smooth = False,
)

3. Heatmap + BBox — Full overlay (recommended for reporting)

result = explainer.display_heatmap_bbox_on_image(
    img_path     = "path/to/image.jpg",
    category     = 1,
    threshold    = 0.5,
    image_weight = 0.5,
    aug_smooth   = False,
    eigen_smooth = False,
)

📊 Visual Results

MS-EFF-VIT — Low-Level Branch

Model	Branch-Level	Image	HiresCam	GradCamElementwise	LayerCam
⚡ ms-eff-vit-b0
🔥 ms-eff-vit-b5

MS-Eff-ViT — High-Level Branch

Model	Branch-Level	Image	EigenGradCam	GradCamPlusPlus	XGradCam
⚡ ms-eff-vit-b0
🔥 ms-eff-vit-b5

MS-EFF-GCVIT — Low-Level Branch

Model	Branch-Level	Image	HiresCam	GradCamElementwise	LayerCam
⚡ ms-eff-gcvit-b0
🔥 ms-eff-gcvit-b5

MS-Eff-GCViT — High-Level Branch

Model	Branch-Level	Image	EigenGradCam	GradCamPlusPlus	XGradCam
⚡ ms-eff-gcvit-b0
🔥 ms-eff-gcvit-b5

📓 Tutorials

The jupyter notebooks themselves can be found under the tutorials folder in the git repository.

• Notebook tutorial: Predict DeepFake Image with ImagePredictor
• Notebook tutorial: Predict DeepFake Video with VideoPredictor
• Notebook tutorial: Low-Level Branch XAI Visualization (HiResCAM, GradCAMElementWise, LayerCAM)
• Notebook tutorial: High-Level Branch XAI Visualization (EigenGradCAM, GradCAM++, XGradCAM)
• Notebook tutorial: MS-EffViT Low-Level Branch Explainability
• Notebook tutorial: MS-EffViT High-Level Branch Explainability
• Notebook tutorial: MS-EffGCViT Low-Level Branch Explainability
• Notebook tutorial: MS-EffGCViT High-Level Branch Explainability

📬 Authors

_{Senior Graduation Project — Department of Software, Chungbuk National University (CBNU), Republic of Korea}

📝 Reference

⚖️ License

This project is licensed under the terms of the MIT license.