| --- |
| license: mit |
| tags: |
| - deepfake-detection |
| - video-classification |
| - efficientnet |
| - celeb df v2 |
| pipeline_tag: video-classification |
| widget: |
| - src: https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/thumbnail.png |
| example_title: "Sample Detection" |
| --- |
| |
| # Deepfake Video Classifier |
|
|
| 🎬 **Detect manipulated videos with 95.73% accuracy** |
|
|
| This model analyzes video frames to determine if content is REAL or a DEEPFAKE. It is Trained on Celebdf v2 dataset and it uses efficientnet B-0. (https://www.kaggle.com/datasets/reubensuju/celeb-df-v2) |
| Developed by Sajjal Fatima, a Software Engineering student at Punjab University College of Information & Technology (PUCIT), Lahore, Pakistan. |
| ## 🚀 Quick Start |
|
|
| ```python |
| from model import DeepFakeModel |
| from utils import video_to_tensor |
| |
| # Load model |
| model = DeepFakeModel("ffpp_efficientnet_best.pth") |
| |
| # Process video |
| video_tensor = video_to_tensor("your_video.mp4") |
| result = model.predict(video_tensor) |
| |
| print(f"Prediction: {result['prediction']}") # REAL or FAKE |
| print(f"Confidence: {result['confidence']:.2%}") |
