---
---
license: mit
language: en
pipeline_tag: image-classification
tags:
- deepfake
- computer-vision
- image-classification
- pytorch
---
---
# 🔥 Deepfake Image Detection Model (Ensemble)

## 📌 Overview
This model detects whether an image is **REAL or FAKE (AI-generated)**.

It uses an **ensemble of 3 Vision Transformer (ViT/CLIP-based) models**:
- m1.safetensors
- m2.safetensors
- m3.safetensors

---

## 🧠 Model Details

| Model | Type |
|------|------|
| m1 | ViT / CLIP |
| m2 | ViT |
| m3 | Fine-tuned CLIP |

---

## 🎯 Task
Binary Classification:
- 0 → REAL
- 1 → FAKE

---

## 📥 Input
- Image (RGB)
- Size: 224 × 224

---

## 📤 Output
- Fake probability (0 to 1)

---

## ⚙️ Installation

```bash
pip install torch torchvision timm safetensors pillow