README.md

---
tags:
- vision
- image-classification
- vit
- transformer
- fake-image-detection
license: apache-2.0
datasets:
- ciplab/real-and-fake-face-detection
model-index:
- name: SahAi
  results:
  - task:
      name: Image Classification
      type: image-classification
    dataset:
      name: Real and Fake Face Detection Dataset
      type: ciplab/real-and-fake-face-detection
    metrics:
      - name: Accuracy
        type: accuracy
        value: 99.12
      - name: Precision
        type: precision
        value: 98.95
      - name: Recall
        type: recall
        value: 99.00
---

# SahAi - Enhanced Fake Image Detection Model
**SahAi** is a fine-tuned Vision Transformer (ViT) model designed for fake image localization in social media.

### 🚀 Model Description
- **Base Model:** ViT-B/16  
- **Input Size:** 224x224  
- **Output Classes:** Real (0), Fake (1)  

### 🔥 How to Use
```python
from transformers import ViTForImageClassification, AutoFeatureExtractor
from PIL import Image
import torch

model = ViTForImageClassification.from_pretrained("SahilSha/SahAi")
feature_extractor = AutoFeatureExtractor.from_pretrained("google/vit-base-patch16-224-in21k")

image = Image.open("test_image.jpg").convert("RGB")
inputs = feature_extractor(images=image, return_tensors="pt")

with torch.no_grad():
    outputs = model(**inputs)
    prediction = outputs.logits.argmax(-1).item()

print("Real" if prediction == 0 else "Fake")