|
|
--- |
|
|
license: apache-2.0 |
|
|
pipeline_tag: image-classification |
|
|
library_name: transformers |
|
|
tags: |
|
|
- deep-fake |
|
|
- ViT |
|
|
- detection |
|
|
- Image |
|
|
base_model: |
|
|
- google/vit-base-patch16-224-in21k |
|
|
datasets: |
|
|
- prithivMLmods/OpenDeepfake-Preview |
|
|
language: |
|
|
- en |
|
|
--- |
|
|
|
|
|
 |
|
|
|
|
|
# deepfake-detector-model |
|
|
|
|
|
> deepfake-detector-model is a vision-language model fine-tuned from `google/vit-base-patch16-224-in21k` for binary image classification. It is trained to detect whether an image is fake or real using the *OpenDeepfake-Preview* dataset. The model uses the `ViTForImageClassification` architecture. |
|
|
|
|
|
```py |
|
|
Classification Report: |
|
|
precision recall f1-score support |
|
|
|
|
|
fake 0.9817 0.9748 0.9782 10000 |
|
|
real 0.9750 0.9818 0.9784 9999 |
|
|
|
|
|
accuracy 0.9783 19999 |
|
|
macro avg 0.9783 0.9783 0.9783 19999 |
|
|
weighted avg 0.9783 0.9783 0.9783 19999 |
|
|
``` |
|
|
|
|
|
 |
|
|
|
|
|
--- |
|
|
|
|
|
## Label Space: 2 Classes |
|
|
|
|
|
The model classifies an image as either: |
|
|
|
|
|
``` |
|
|
Class 0: fake |
|
|
Class 1: real |
|
|
``` |
|
|
|
|
|
--- |
|
|
|
|
|
## Install Dependencies |
|
|
|
|
|
```bash |
|
|
pip install -q transformers torch pillow gradio hf_xet |
|
|
``` |
|
|
|
|
|
--- |
|
|
|
|
|
## Inference Code |
|
|
|
|
|
```python |
|
|
import gradio as gr |
|
|
from transformers import ViTImageProcessor, ViTForImageClassification |
|
|
from PIL import Image |
|
|
import torch |
|
|
|
|
|
# Load model and processor |
|
|
model_name = "prithivMLmods/deepfake-detector-model" |
|
|
model = ViTForImageClassification.from_pretrained(model_name) |
|
|
processor = ViTImageProcessor.from_pretrained(model_name) |
|
|
|
|
|
# Updated label mapping |
|
|
labels_list = ['fake', 'real'] |
|
|
|
|
|
def classify_image(image): |
|
|
image = Image.fromarray(image).convert("RGB") |
|
|
inputs = processor(images=image, return_tensors="pt") |
|
|
|
|
|
with torch.no_grad(): |
|
|
outputs = model(**inputs) |
|
|
logits = outputs.logits |
|
|
probs = torch.nn.functional.softmax(logits, dim=1).squeeze().tolist() |
|
|
|
|
|
prediction = { |
|
|
labels_list[i]: round(probs[i], 3) for i in range(len(probs)) |
|
|
} |
|
|
|
|
|
return prediction |
|
|
|
|
|
# Gradio Interface |
|
|
iface = gr.Interface( |
|
|
fn=classify_image, |
|
|
inputs=gr.Image(type="numpy"), |
|
|
outputs=gr.Label(num_top_classes=2, label="Deepfake Detection"), |
|
|
title="deepfake-detector-model", |
|
|
description="Upload an image to detect whether it is AI-generated (fake) or a real photograph (real), using the OpenDeepfake-Preview dataset." |
|
|
) |
|
|
|
|
|
if __name__ == "__main__": |
|
|
iface.launch() |
|
|
``` |
|
|
|
|
|
--- |
|
|
|
|
|
## Intended Use |
|
|
|
|
|
`deepfake-detector-model` is designed for: |
|
|
|
|
|
* **Deepfake Detection** – Identify AI-generated or manipulated images. |
|
|
* **Content Moderation** – Flag synthetic or fake visual content. |
|
|
* **Dataset Curation** – Remove synthetic samples from mixed datasets. |
|
|
* **Visual Authenticity Verification** – Check the integrity of visual media. |
|
|
* **Digital Forensics** – Support image source verification and traceability. |
