This project is about using Pytorch for 2D MRI image classification using transfer learning with Resnet34 with a testing accuracy is 98%.
The images are downloaded from Kaggle. These images will be classified into four categories, including
NOD: Non Demented,
VMD: Very Mild Demented,
MID: Mild Demented,
MOD: Moderate Demented.
| Index | Description | Jupiter notebook | Content | data |
|---|---|---|---|---|
| 1 | Val_accuracy: 0.98 Testing accuracy: 0.68 😢 |
01_Resrnet34.ipynb | Download data split to train, val, test Train with Resnet34 Testing evaluation Reasoning |
org_day |
| 2 | Val_accuracy: 0.99 Testing accuracy: 0.98 😃 Is this approach ok? 🤔 |
02_Resnet34.ipynb | Combine data Split data Train model Evaluate the model |
allnew |
For utility functions, please see mymodulo.py
- 01-Resnet34.ipynb, 02-Resnet34.ipynb: Two main run files.
- mymodule.py: all utility functions used in the project.
- model: folder contains trained model.
- Data: contains data after being downloaded.