| --- |
| library_name: keras |
| pipeline_tag: image-classification |
| license: mit |
| tags: |
| - deepfake-detection |
| - xception |
| - tensorflow |
| - keras |
| - computer-vision |
| --- |
| |
| # DeepVerify Xception Model |
|
|
| DeepVerify is a deepfake face detection model developed using Xception Transfer Learning and TensorFlow/Keras. |
|
|
| The model is trained to classify facial images into: |
|
|
| - Real Face |
| - Fake Face |
|
|
| It is designed for AI-powered media verification, deepfake detection, and facial authenticity analysis. |
|
|
| ## Model Details |
|
|
| - Architecture: Xception |
| - Framework: TensorFlow / Keras |
| - Input Size: 299 × 299 RGB |
| - Task: Deepfake Face Detection |
|
|
| ### Output Classes |
|
|
| | Class | Label | |
| |---------|---------| |
| | 0 | Fake Face | |
| | 1 | Real Face | |
|
|
| ## Performance |
|
|
| - Test Accuracy: 91.37% |
| - Binary Classification |
|
|
| ## Usage |
|
|
|
|
| from tensorflow.keras.models import load_model |
| |
| model = load_model("best_xception.keras") |
| |
| ## Author |
| |
| **Naveen** |
| |
| Artificial Intelligence & Data Science |
| Velammal Institute of Technology |
| |
| Project: DeepVerify AI |
