Adds gradio app
Browse files
app.py
ADDED
|
@@ -0,0 +1,50 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import torch
|
| 2 |
+
import gradio as gr
|
| 3 |
+
from src.model import DRModel
|
| 4 |
+
from torchvision import transforms as T
|
| 5 |
+
|
| 6 |
+
CHECKPOINT_PATH = "checkpoints/epoch=19-step=8800.ckpt"
|
| 7 |
+
model = DRModel.load_from_checkpoint(CHECKPOINT_PATH)
|
| 8 |
+
|
| 9 |
+
labels = {
|
| 10 |
+
0: "No DR",
|
| 11 |
+
1: "Mild",
|
| 12 |
+
2: "Moderate",
|
| 13 |
+
3: "Severe",
|
| 14 |
+
4: "Proliferative DR",
|
| 15 |
+
}
|
| 16 |
+
|
| 17 |
+
transform = T.Compose(
|
| 18 |
+
[
|
| 19 |
+
T.Resize((192, 192)),
|
| 20 |
+
T.ToTensor(),
|
| 21 |
+
T.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
|
| 22 |
+
]
|
| 23 |
+
)
|
| 24 |
+
|
| 25 |
+
# Define the prediction function
|
| 26 |
+
def predict(input_img):
|
| 27 |
+
input_img = transform(input_img).unsqueeze(0)
|
| 28 |
+
with torch.no_grad():
|
| 29 |
+
prediction = torch.nn.functional.softmax(model(input_img)[0], dim=0)
|
| 30 |
+
confidences = {labels[i]: float(prediction[i]) for i in labels}
|
| 31 |
+
return confidences
|
| 32 |
+
|
| 33 |
+
|
| 34 |
+
# Set up the Gradio app interface
|
| 35 |
+
dr_app = gr.Interface(
|
| 36 |
+
fn=predict,
|
| 37 |
+
inputs=gr.Image(type="pil"),
|
| 38 |
+
outputs=gr.Label(),
|
| 39 |
+
title="Diabetic Retinopathy Detection",
|
| 40 |
+
examples=[
|
| 41 |
+
"data/sample/10_left.jpeg",
|
| 42 |
+
"data/sample/10_right.jpeg",
|
| 43 |
+
"data/sample/15_left.jpeg",
|
| 44 |
+
"data/sample/16_right.jpeg",
|
| 45 |
+
],
|
| 46 |
+
)
|
| 47 |
+
|
| 48 |
+
# Run the Gradio app
|
| 49 |
+
if __name__ == "__main__":
|
| 50 |
+
dr_app.launch()
|