karolprando/deepfake-detector-v1

Model Card for Model ID

Modelo com fine-tuning para TCC, treinado com aproximadamente 140 mil imagens.

Model Details

Model Description

• Model type: Classification
• Language(s) (NLP): ViT
• Finetuned from model: https://huggingface.co/dima806/deepfake_vs_real_image_detection

Model Sources

• Repository: [More Information Needed]

Uses

O modelo foi retreinado para fins academicos, como parte da disciplina de TCC.

How to Get Started with the Model

from transformers import AutoModelForImageClassification, AutoImageProcessor
from PIL import Image
import torch

model_path = 'karolprando/deepfake-detector-v1'

model = AutoModelForImageClassification.from_pretrained(model_path)
processor = AutoImageProcessor.from_pretrained(model_path)

img_path = r"C:\Users\Karol\OneDrive\Imagens\popeia.jpeg" # substitui aqui
image = Image.open(img_path)

inputs = processor(images=image, return_tensors="pt")
with torch.no_grad():
    outputs = model(**inputs)
    preds = torch.nn.functional.softmax(outputs.logits, dim=-1)
    predicted_label = preds.argmax(-1).item()

print("Predição:", model.config.id2label[predicted_label])

Training Details

Training Data

Link dataset: https://www.kaggle.com/datasets/manjilkarki/deepfake-and-real-images

Metrics

[More Information Needed]