fish_freshness_app.py

import torch
import torch.nn.functional as F
from torchvision import transforms
from PIL import Image
import gradio as gr
import numpy as np
import cv2

# Load the model class definition
from models.efficientnet_b0 import EfficientNetB0Classifier

# Constants
MODEL_PATH = "efficientnet_best9912.pth"
CLASS_NAMES = ["Fresh", "Not Fresh"]
INPUT_SIZE = 380
MODEL_ACCURACY = "99.12%"  # Your model's validation accuracy

# Define preprocessing pipeline
preprocess = transforms.Compose([
    transforms.Resize((INPUT_SIZE, INPUT_SIZE)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406],
                       std=[0.229, 0.224, 0.225])
])

# Load model
def load_model():
    model = EfficientNetB0Classifier(train_base=False)
    model.load_state_dict(torch.load(MODEL_PATH, map_location=torch.device('cpu')))
    model.eval()
    return model

model = load_model()

def process_prediction(confidence_score):
    """Convert model output to detailed prediction information"""
    fresh_prob = float(confidence_score)
    not_fresh_prob = float(1.0 - confidence_score)
    
    prediction = "Fresh" if fresh_prob > 0.5 else "Not Fresh"
    confidence = fresh_prob if fresh_prob > 0.5 else not_fresh_prob
    
    return {
        "Fresh": fresh_prob,
        "Not Fresh": not_fresh_prob
    }, prediction, confidence

def analyze_image(image):
    """Analyze the image and return detailed results"""
    if image is None:
        return None, None, None, None
    
    # Convert to RGB if needed
    if len(image.shape) == 2:  # Grayscale
        image = cv2.cvtColor(image, cv2.COLOR_GRAY2RGB)
    elif image.shape[2] == 4:  # RGBA
        image = cv2.cvtColor(image, cv2.COLOR_RGBA2RGB)
    
    # Prepare image for model
    pil_image = Image.fromarray(image).convert('RGB')
    input_tensor = preprocess(pil_image).unsqueeze(0)
    
    # Get prediction
    with torch.no_grad():
        output = model(input_tensor)
        confidence_score = output.item()
    
    # Process results
    probabilities, prediction, confidence = process_prediction(confidence_score)
    
    # Create result message
    confidence_percentage = f"{confidence * 100:.2f}%"
    message = f"Prediction: {prediction} (Confidence: {confidence_percentage})"
    
    # Prepare visualization
    display_image = cv2.resize(image, (INPUT_SIZE, INPUT_SIZE))
    
    return probabilities, message, display_image, confidence_percentage

# Custom CSS for better styling
custom_css = """
.gradio-container {
    font-family: 'IBM Plex Sans', sans-serif;
}
.gr-button {
    color: white;
    border-radius: 8px;
    background: linear-gradient(45deg, #4CAF50, #45a049);
    border: none;
    font-size: 1.2em;
    padding: 10px 20px;
}
.gr-button:hover {
    background: linear-gradient(45deg, #45a049, #4CAF50);
    transform: translateY(-2px);
    box-shadow: 0 5px 15px rgba(0,0,0,0.1);
}
.footer {
    margin-top: 20px;
    text-align: center;
    font-size: 0.8em;
}
.confidence {
    font-size: 1.2em;
    font-weight: bold;
    margin-top: 10px;
}
.container {
    max-width: 1200px;
    margin: 0 auto;
    padding: 20px;
}
.result-box {
    background: #f8f9fa;
    border-radius: 10px;
    padding: 20px;
    margin-top: 20px;
    box-shadow: 0 2px 10px rgba(0,0,0,0.1);
}
"""

# Create Gradio interface
with gr.Blocks(css=custom_css) as demo:
    gr.Markdown(
        """
        # 🐟 Fish Freshness Classifier
        
        Upload a fish image and get instant freshness analysis using our advanced AI model.
        
        ### Model Performance
        - Architecture: EfficientNet-B0
        - Validation Accuracy: """ + MODEL_ACCURACY + """
        """
    )
    
    with gr.Row():
        with gr.Column(scale=1):
            input_image = gr.Image(
                label="Upload Fish Image",
                type="numpy",
                height=400,
                sources=["upload", "webcam", "clipboard"]
            )
            upload_button = gr.Button("📸 Analyze Freshness", variant="primary", size="lg")
        
        with gr.Column(scale=1):
            with gr.Group(elem_classes="result-box"):
                output_label = gr.Label(
                    num_top_classes=2,
                    label="Freshness Analysis",
                    show_label=True
                )
                result_message = gr.Textbox(
                    label="Detailed Result",
                    show_copy_button=True
                )
                confidence_indicator = gr.Textbox(
                    label="Confidence Level",
                    show_copy_button=True
                )
    
    gr.Markdown(
        """
        ### 📝 Best Practices
        - Use clear, well-lit images
        - Ensure the fish is clearly visible
        - Include key features (eyes, gills, skin)
        - Avoid blurry or dark photos
        """
    )
    
    # Set up the prediction flow
    upload_button.click(
        fn=analyze_image,
        inputs=input_image,
        outputs=[output_label, result_message, input_image, confidence_indicator]
    )

if __name__ == "__main__":
    demo.launch(share=True)