Create app.py

app.py

@@ -0,0 +1,36 @@
+import gradio as gr
+import wandb
+def predict_image(image):
+    image = Image.fromarray(image).convert("RGB")
+    image = transform(image)
+    image = image.unsqueeze(0)  # Add batch dimension
+
+    with torch.no_grad():
+        outputs = model(image)
+        probabilities = F.softmax(outputs, dim=1)
+        _, predicted = torch.max(outputs, 1)
+
+    predicted_label = classes[predicted.item()]
+    predicted_probability = probabilities[0][predicted.item()].item()
+
+    # Log to W&B
+    wandb.log({"image": [wandb.Image(image, caption="Input Image")],
+               "predicted_label": predicted_label,
+               "predicted_probability": predicted_probability})
+
+    return predicted_label, predicted_probability
+
+# Class labels
+classes = ['dew', 'fog_smog', 'frost', 'glaze', 'hail', 'lightning', 'rain', 'rainbow', 'rime', 'sandstorm', 'snow']
+
+# Create Gradio interface
+interface = gr.Interface(fn=predict_image, 
+                         inputs=gr.components.Image(), 
+                         outputs=[gr.components.Textbox(label="Predicted Label"), gr.components.Textbox(label="Prediction Probability")],title="Weather Detection App")
+
+# Launch the interface
+interface.launch()
+
+# Save model checkpoint to W&B
+torch.save(model.state_dict(), 'weather_model.pth')
+wandb.save('weather_model.pth')