app.py

### 1. Imports and class names setup ### 
import gradio as gr
import os
import torch

from torch import nn
from model import create_resnet50_model
from timeit import default_timer as timer
from typing import Tuple, Dict

import torch.nn.functional as F

# Setup class names
class_names = ['CRVO',
  'Choroidal Nevus',
  'Diabetic Retinopathy',
  'Laser Spots',
  'Macular Degeneration',
  'Macular Hole',
  'Myelinated Nerve Fiber',
  'Normal',
  'Pathological Mypoia',
  'Retinitis Pigmentosa']

### 2. Model and transforms preparation ###

# Create ResNet50 model
resnet50, resnet50_transforms = create_resnet50_model(
    num_classes=len(class_names), # actual value would also work
)
resnet50.fc = nn.Linear(2048, 10)

# Load saved weights
resnet50.load_state_dict(
    torch.load(
        f="pretrained_resnet50_feature_extractor_drappcompressed.pth",
        map_location=torch.device("cpu"),  # load to CPU
    )
)


### 3. Predict function ###

# Create predict function
# def predict(img) -> Tuple[Dict, float]:
#     """Transforms and performs a prediction on img and returns prediction and time taken.
#     """
#     # Start the timer
#     start_time = timer()
    
#     # Transform the target image and add a batch dimension
#     img = resnet50_transforms(img).unsqueeze(0)
    
#     # Put model into evaluation mode and turn on inference mode
#     resnet50.eval()
#     with torch.inference_mode():
#         # Pass the transformed image through the model and turn the prediction logits into prediction probabilities
#         pred_probs = torch.softmax(resnet50(img), dim=1)
    
#     # Create a prediction label and prediction probability dictionary for each prediction class (this is the required format for Gradio's output parameter)
#     pred_labels_and_probs = {class_names[i]: float(pred_probs[0][i]) for i in range(len(class_names))}
    
#     # Calculate the prediction time
#     pred_time = round(timer() - start_time, 5)
    
#     # Return the prediction dictionary and prediction time 
#     return pred_labels_and_probs, pred_time

def predict(img):
    """Transforms and performs a prediction on img and returns prediction and time taken."""
    start_time = timer()
    
    try:
        img = resnet50_transforms(img).unsqueeze(0)
        resnet50.eval()
        
        with torch.inference_mode():
            pred_probs = torch.softmax(resnet50(img), dim=1)
            
            # Calculate entropy for OOD detection
            entropy = -torch.sum(pred_probs * torch.log(pred_probs + 1e-8)).item()
            max_prob = torch.max(pred_probs).item()
            
            # Create base prediction dictionary
            pred_labels_and_probs = {class_names[i]: float(pred_probs[0][i]) for i in range(len(class_names))}
            
            # OOD Detection - modify existing probabilities instead of adding new keys
            if (max_prob > 0.95 and entropy < 0.2) or entropy > 2.0:
                # Boost the probability of the first class and add a marker
                pred_labels_and_probs[class_names[0]] = 0.99  # Use existing class
                # You could also just print a warning or log it
                print("May not be retina scan")
        
        pred_time = round(timer() - start_time, 5)
        return pred_labels_and_probs, pred_time
        
    except Exception as e:
        # Return dictionary with same structure as normal case
        pred_labels_and_probs = {class_names[i]: 0.0 for i in range(len(class_names))}
        pred_labels_and_probs[class_names[0]] = 1.0  # Show error in first class
        return pred_labels_and_probs, 0.0

### 4. Gradio app ###

# Create title, description and article strings
#title = "DeepFundus 👀"
#description = "A ResNet50 feature extractor computer vision model to classify funduscopic images."
#article = "Created with the help from [09. PyTorch Model Deployment](https://www.learnpytorch.io/09_pytorch_model_deployment/)."

# Create examples list from "examples/" directory
example_list = [["examples/" + example] for example in os.listdir("examples")]

# Create the Gradio demo
demo = gr.Interface(fn=predict, # mapping function from input to output
                    inputs=gr.Image(type="pil"), # what are the inputs?
                    outputs=[gr.Label(num_top_classes=3, label="Predictions"), # what are the outputs?
                             gr.Number(label="Prediction time (s)")], # our fn has two outputs, therefore we have two outputs
                    # Create examples list from "examples/" directory
                    examples=example_list) 
                    #title=title,
                    #description=description,
                    #article=article)

# Launch the demo!
demo.launch()