app.py

import gradio as gr
import torch
import torch.nn as nn
from torchvision import transforms
import torchvision.models as models
from PIL import Image
import numpy as np

# Function to load class names from a file
def load_class_names(file_path):
    with open(file_path, 'r') as f:
        class_names = [line.strip() for line in f.readlines()]
    return class_names

# Function to load Wikipedia links from a file
def load_class_info(file_path):
    with open(file_path, 'r') as f:
        class_info = [line.strip() for line in f.readlines()]
    return class_info

# Function to load the model from a .pkl file
def load_model(model_path, model_type):
    # Load the model state dictionary
    model_state_dict = torch.load(model_path, map_location=torch.device('cpu'))

    # Create an instance of the model based on model_type
    if model_type == 'mobilenetv2':
        model = models.mobilenet_v2(pretrained=False)
        model.classifier[1] = nn.Linear(model.last_channel, num_classes)
    elif model_type == 'resnet18':
        model = models.resnet18(pretrained=False)
        model.fc = nn.Linear(model.fc.in_features, num_classes)
    elif model_type == 'densenet121':
        model = models.densenet121(pretrained=False)
        model.classifier = nn.Linear(model.classifier.in_features, num_classes)
    else:
        raise ValueError(f"Unsupported model type: {model_type}")

    # Load the state dictionary into the model
    model.load_state_dict(model_state_dict)
    # Set the model to evaluation mode
    model.eval()
    return model

# Define the transformation
val_transform = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])

# Define the prediction function
def predict_image(image, model_choice):
    global model, current_model

    # Check if a model is selected
    if model_choice not in model_paths:
        return "Error: Please select a valid model.", ""

    # Load the selected model if it's not already loaded
    if model_choice != current_model:
        model_path = model_paths[model_choice]
        model = load_model(model_path, model_choice)
        current_model = model_choice
    # Convert the NumPy array to a PIL Image if needed
    if isinstance(image, np.ndarray):
        image = Image.fromarray(image.astype('uint8'), 'RGB')
   
    image = val_transform(image)
    image = image.unsqueeze(0)  # Add batch dimension

    with torch.no_grad():
        outputs = model(image)
        probabilities = torch.nn.functional.softmax(outputs[0], dim=0)
        confidence, preds = torch.max(probabilities, 0)

    confidence_score = confidence.item() * 100

    if confidence_score < 30:
        result = "Not identified"
        html_result = ""
    else:
        class_name = class_names[preds.item()]
        wiki_link = class_info[preds.item()]
        result = f"{class_name}: {confidence_score:.2f}%"
        html_result = f"<h1><br><a href='{wiki_link}' target='_blank'>More Info</a></h1>"

    return result, html_result

# Load class names and class info
class_file_path = 'classes.txt'
class_info_path = 'classinfo.txt'
class_names = load_class_names(class_file_path)
class_info = load_class_info(class_info_path)
num_classes = len(class_names)

# Define model paths
model_paths = {
    'densenet121': 'densenet121_15EpochsPretrainedNoExtractionNoLR_model.pkl',
    'resnet18': 'resnet18_25EpochsPretrainedExtractionNoLR_model.pkl',
    'mobilenetv2': 'mobilenetv2_25EpochsPretrainedExtractionNoLR_model.pkl'
}

# Set default model
current_model = 'densenet121'
model = load_model(model_paths[current_model], current_model)

# Create the Gradio interface
iface = gr.Interface(
    fn=predict_image,
    inputs=[
        gr.Image(height=500),
        gr.Dropdown(choices=["densenet121", "resnet18", "mobilenetv2"], value="densenet121", label="Select Model")
    ],
    outputs=[gr.Label(num_top_classes=1), gr.HTML()],
    title="Animal Classification",
    description="Upload an image to get the predicted label",
    allow_flagging="never",
)

# Launch the interface
iface.launch()