|
|
--- |
|
|
metrics: |
|
|
- accuracy: 0.9815 |
|
|
- Intersection Over Union: 0.9509 |
|
|
- Dice score: 0.9794 |
|
|
datasets: |
|
|
- tangezerman/ain3007 |
|
|
base_model: |
|
|
- 5mp-hub/vgg16.imagenet |
|
|
--- |
|
|
|
|
|
# VGG16-U-NET Model Card |
|
|
|
|
|
## Model Description |
|
|
|
|
|
**Model Name:** VGG16-U-NET |
|
|
|
|
|
**Model Type:** Image Segmentation |
|
|
|
|
|
**Architecture:** U-Net with VGG16 weights trained on Imagenet |
|
|
|
|
|
|
|
|
## Model Performance |
|
|
|
|
|
| Metric | Value | |
|
|
|--------|-------| |
|
|
| Accuracy | 0.9815 | |
|
|
| Intersection Over Union (IoU) | 0.9509 | |
|
|
| Dice Score | 0.9794 | |
|
|
|
|
|
## Training Details |
|
|
|
|
|
**Dataset:** tangez/ain3007 |
|
|
|
|
|
**Training Parameters:** |
|
|
- Architecture: U-Net with VGG16 encoder backbone |
|
|
- Pre-trained weights: ImageNet |
|
|
- Framework: PyTorch |
|
|
|
|
|
## Intended Use |
|
|
|
|
|
**Primary Use Cases:** |
|
|
- Image segmentation tasks |
|
|
|
|
|
## How to Use |
|
|
|
|
|
```python |
|
|
import torch |
|
|
|
|
|
# Load the trained model from the Models directory |
|
|
model = torch.load("path/to/model.pth") |
|
|
model.eval() |
|
|
|
|
|
# For inference on WSI patches |
|
|
with torch.no_grad(): |
|
|
output = model(input_patches) |
|
|
# Output will be binary tissue masks |
|
|
``` |
|
|
|
