| license: mit | |
| tags: | |
| - image-classification | |
| - fraud-detection | |
| - cheque-verification | |
| - onnx | |
| - pytorch | |
| - computer-vision | |
| metrics: | |
| - accuracy | |
| # Aegis ChequeGuard: Real vs. Fake Cheque Verification Engine 🛡️ | |
| An advanced, production-ready computer vision model designed to classify and authenticate bank cheques as **Real** or **Fake/Photocopy** with high precision. | |
| ## 🚀 Model Details | |
| - **Architecture:** ResNet50 (Transfer Learning with custom classification head) | |
| - **Task:** Binary Image Classification (Real Cheque vs. Fake Cheque) | |
| - **Formats Available:** PyTorch (`.pth`) and Cross-Platform Optimized (`.onnx`) | |
| - **Training Data:** Augmented financial document dataset optimized for real-world variations. | |
| ## 📂 Files Included | |
| - `best_augmented_cheque_model.pth`: PyTorch trained model weights. | |
| - `cheque_verification_model.onnx`: Exported ONNX model optimized for edge devices, mobile apps, and fast CPU/GPU inference. | |