app.py

"""
Poultry Disease Classification System
Professional diagnostic tool using EfficientNetV2
"""

import gradio as gr
import numpy as np
import tensorflow as tf
from PIL import Image
import keras

# Load model
MODEL_PATH = "poultry_disease_classification_efficientnet_model.keras"
print("Loading model...")
try:
    model = keras.models.load_model(MODEL_PATH)
    print("Model loaded successfully")
except Exception as e:
    print(f"Error loading model: {e}")
    raise

# Class definitions
CLASS_NAMES = ['Coccidiosis', 'Healthy', 'Newcastle Disease', 'Salmonella']

# Disease information - clinical and concise
DISEASE_INFO = {
    'Coccidiosis': {
        'description': 'Parasitic infection of the intestinal tract caused by coccidia protozoa.',
        'symptoms': [
            'Bloody or watery droppings',
            'Lethargy and weakness',
            'Reduced appetite',
            'Pale comb and wattles'
        ],
        'treatment': [
            'Anticoccidial medication (Amprolium, Sulfonamides)',
            'Electrolyte supplementation',
            'Isolation of affected birds',
            'Veterinary consultation required'
        ],
        'prevention': [
            'Maintain dry, clean housing',
            'Regular disinfection protocols',
            'Adequate ventilation'
        ]
    },
    'Healthy': {
        'description': 'No signs of disease detected. Bird appears to be in good health.',
        'symptoms': [
            'Alert and active behavior',
            'Normal appetite and droppings',
            'Clear eyes',
            'Well-groomed feathers'
        ],
        'treatment': [
            'Continue standard care protocols',
            'Monitor regularly',
            'Maintain balanced nutrition'
        ],
        'prevention': [
            'Balanced feed formulation',
            'Clean environment',
            'Vaccination schedule'
        ]
    },
    'Newcastle Disease': {
        'description': 'Highly contagious viral infection affecting respiratory, nervous, and digestive systems.',
        'symptoms': [
            'Respiratory distress',
            'Neurological signs (tremors, paralysis)',
            'Green watery diarrhea',
            'Reduced egg production',
            'Head and neck swelling'
        ],
        'treatment': [
            'No specific antiviral treatment available',
            'Supportive care and isolation',
            'Vaccination of unaffected birds',
            'Strict biosecurity enforcement'
        ],
        'prevention': [
            'Vaccination program (primary prevention)',
            'Biosecurity protocols',
            'Quarantine procedures for new birds'
        ]
    },
    'Salmonella': {
        'description': 'Bacterial infection with zoonotic potential. Affects gastrointestinal system.',
        'symptoms': [
            'Greenish diarrhea',
            'Decreased appetite',
            'Dehydration',
            'Lethargy',
            'Poor growth in young birds'
        ],
        'treatment': [
            'Antibiotic therapy (veterinary prescribed)',
            'Fluid and electrolyte therapy',
            'Probiotic supplementation',
            'Enhanced sanitation measures'
        ],
        'prevention': [
            'Biosecurity compliance',
            'Water and feed quality control',
            'Proper waste management'
        ]
    }
}

def preprocess_image(image):
    """Preprocess image for model input"""
    if isinstance(image, np.ndarray):
        image = Image.fromarray(image.astype('uint8'))
    
    image = image.resize((360, 360))
    img_array = np.array(image)
    img_array = img_array.astype('float32') / 255.0
    img_array = np.expand_dims(img_array, axis=0)
    
    return img_array

def predict_disease(image):
    """Generate disease prediction and analysis"""
    if image is None:
        return None, "<div class='error-msg'>Please upload an image to begin analysis</div>"
    
    try:
        processed_image = preprocess_image(image)
        predictions = model.predict(processed_image, verbose=0)[0]
        
        predicted_idx = np.argmax(predictions)
        predicted_class = CLASS_NAMES[predicted_idx]
        confidence = predictions[predicted_idx] * 100
        
        # Strict threshold - only proceed if confidence >= 99%
        CONFIDENCE_THRESHOLD = 98.7
        
        if confidence < CONFIDENCE_THRESHOLD:
            # Simple rejection message - NO analysis details
            error_html = """
            <div class='error-container'>
                <strong>⚠️ Invalid Image Detected</strong><br><br>
                The uploaded image does not appear to be a poultry bird or is not suitable for analysis.<br><br>
                
                <strong>Please upload:</strong>
                <ul style='margin: 0.5rem 0; padding-left: 1.5rem; line-height: 1.6;'>
                    <li>A clear photograph of a chicken or poultry bird</li>
                    <li>Well-lit image with good focus</li>
                    <li>Bird clearly visible in the frame</li>
                    <li>No other animals or unrelated objects</li>
                </ul>
                
                <p style='margin-top: 1rem; font-size: 0.85rem; color: #6c757d;'>
                This system is specifically trained to classify poultry diseases and requires 
                high-confidence detections to ensure diagnostic accuracy.
                </p>
            </div>
            """
            return None, error_html
        
        # Proceed with normal analysis for high-confidence predictions (≥99%)
        prediction_dict = {CLASS_NAMES[i]: float(predictions[i]) for i in range(len(CLASS_NAMES))}
        disease_info = DISEASE_INFO[predicted_class]
        
        # Confidence assessment (for valid predictions only)
        if confidence >= 99.5:
            conf_level = "Very High"
            conf_class = "conf-high"
        elif confidence >= 99.0:
            conf_level = "High"
            conf_class = "conf-high"
        else:
            # This shouldn't happen, but keeping as fallback
            conf_level = "Moderate"
            conf_class = "conf-moderate"
        
        # Build clean, professional result display
        result_html = f"""
        <div class='result-container'>
            <div class='result-header'>
                <div class='diagnosis'>
                    <span class='label'>Diagnosis</span>
                    <span class='value'>{predicted_class}</span>
                </div>
                <div class='confidence {conf_class}'>
                    <span class='label'>Confidence</span>
                    <span class='value'>{confidence:.2f}%</span>
                    <span class='level'>{conf_level}</span>
                </div>
            </div>
            
            <div class='info-section'>
                <h3>Clinical Overview</h3>
                <p>{disease_info['description']}</p>
            </div>
            
            <div class='info-section'>
                <h3>Key Symptoms</h3>
                <ul>
                    {''.join([f'<li>{symptom}</li>' for symptom in disease_info['symptoms']])}
                </ul>
            </div>
            
            <div class='info-section'>
                <h3>Treatment Protocol</h3>
                <ul>
                    {''.join([f'<li>{treatment}</li>' for treatment in disease_info['treatment']])}
                </ul>
            </div>
            
            <div class='info-section'>
                <h3>Prevention</h3>
                <ul>
                    {''.join([f'<li>{prevention}</li>' for prevention in disease_info['prevention']])}
                </ul>
            </div>
            
            <div class='disclaimer'>
                <strong>Professional Consultation Required</strong><br>
                This AI diagnostic tool provides preliminary screening only. 
                Consult a licensed veterinarian for definitive diagnosis and treatment.
            </div>
        </div>
        """
        
        return prediction_dict, result_html
        
    except Exception as e:
        error_html = f"""
        <div class='error-container'>
            <strong>Analysis Error</strong><br>
            {str(e)}<br><br>
            Please verify the image and try again.
        </div>
        """
        return None, error_html
        
# Professional CSS styling
custom_css = """
* {
    font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Oxygen, Ubuntu, Cantarell, sans-serif;
}

.gradio-container {
    max-width: 1200px !important;
    margin: 0 auto !important;
}

.header {
    background: linear-gradient(135deg, #1e3c72 0%, #2a5298 100%);
    color: white !important;
    padding: 2.5rem 2rem;
    border-radius: 8px;
    margin-bottom: 2rem;
}

.header h1 {
    font-size: 1.75rem;
    font-weight: 600;
    margin: 0 0 0.5rem 0;
    letter-spacing: -0.02em;
    color: white !important;
}

.header p {
    font-size: 0.95rem;
    margin: 0;
    opacity: 0.95;
    font-weight: 400;
    color: white !important;
}

.classes-grid {
    display: grid;
    grid-template-columns: repeat(4, 1fr);
    gap: 1rem;
    margin: 2rem 0;
    padding: 0 0.5rem;
}

.class-card {
    background: #f8f9fa;
    padding: 1rem;
    border-radius: 6px;
    text-align: center;
    border: 1px solid #e9ecef;
}

.class-card h4 {
    font-size: 0.9rem;
    font-weight: 600;
    color: #212529;
    margin: 0 0 0.25rem 0;
}

.class-card p {
    font-size: 0.8rem;
    color: #6c757d;
    margin: 0;
}

/* Dark mode support */
@media (prefers-color-scheme: dark) {
    .class-card {
        background: #2d3748;
        border: 1px solid #4a5568;
    }
    
    .class-card h4 {
        color: #e2e8f0 !important;
    }
    
    .class-card p {
        color: #cbd5e0 !important;
    }
}

/* Hide the label/title above upload section */
.upload-section label.block {
    display: none !important;
}

.upload-section .gr-block {
    border: none !important;
    padding: 0 !important;
}

.upload-section {
    background: transparent;
    padding: 0;
    border: none;
}

.section-title {
    font-size: 1rem;
    font-weight: 600;
    color: #212529;
    margin: 0 0 1rem 0;
    background: white;
    padding: 1rem;
    border-radius: 8px 8px 0 0;
    border: 1px solid #e9ecef;
    border-bottom: none;
}

/* Dark mode for section title */
@media (prefers-color-scheme: dark) {
    .section-title {
        background: #1a202c;
        color: #e2e8f0 !important;
        border-color: #4a5568;
    }
}

.image-tips {
    background: #f8f9fa;
    padding: 1rem;
    border-radius: 6px;
    margin-top: 1rem;
    border-left: 3px solid #2a5298;
}

.image-tips h4 {
    font-size: 0.85rem;
    font-weight: 600;
    color: #212529;
    margin: 0 0 0.5rem 0;
}

.image-tips ul {
    margin: 0;
    padding-left: 1.25rem;
    font-size: 0.85rem;
    color: #495057;
    line-height: 1.6;
}

/* Dark mode for image tips */
@media (prefers-color-scheme: dark) {
    .image-tips {
        background: #2d3748;
        border-left-color: #4299e1;
    }
    
    .image-tips h4 {
        color: #e2e8f0 !important;
    }
    
    .image-tips ul {
        color: #cbd5e0 !important;
    }
}

.result-container {
    background: #ffffff;
    border-radius: 8px;
    overflow: hidden;
    border: 1px solid #e9ecef;
}

/* Dark mode for results */
@media (prefers-color-scheme: dark) {
    .result-container {
        background: #1a202c;
        border-color: #4a5568;
    }
}

.result-header {
    display: grid;
    grid-template-columns: 1fr 1fr;
    gap: 1rem;
    padding: 1.5rem;
    background: #f8f9fa;
    border-bottom: 1px solid #e9ecef;
}

/* Dark mode for result header */
@media (prefers-color-scheme: dark) {
    .result-header {
        background: #2d3748;
        border-bottom-color: #4a5568;
    }
}

.diagnosis, .confidence {
    display: flex;
    flex-direction: column;
}

.label {
    font-size: 0.75rem;
    text-transform: uppercase;
    letter-spacing: 0.05em;
    color: #6c757d;
    font-weight: 600;
    margin-bottom: 0.25rem;
}

.value {
    font-size: 1.5rem;
    font-weight: 600;
    color: #212529;
}

.level {
    font-size: 0.85rem;
    color: #6c757d;
    margin-top: 0.25rem;
}

/* Dark mode for labels and values */
@media (prefers-color-scheme: dark) {
    .label {
        color: #a0aec0 !important;
    }
    
    .value {
        color: #e2e8f0 !important;
    }
    
    .level {
        color: #a0aec0 !important;
    }
}

.conf-high .value { color: #28a745 !important; }
.conf-moderate .value { color: #ffc107 !important; }
.conf-low .value { color: #dc3545 !important; }

.info-section {
    padding: 1.5rem;
    border-bottom: 1px solid #e9ecef;
}

.info-section:last-of-type {
    border-bottom: none;
}

.info-section h3 {
    font-size: 0.95rem;
    font-weight: 600;
    color: #212529;
    margin: 0 0 0.75rem 0;
}

.info-section p {
    font-size: 0.9rem;
    color: #495057;
    line-height: 1.6;
    margin: 0;
}

.info-section ul {
    margin: 0;
    padding-left: 1.25rem;
    font-size: 0.9rem;
    color: #495057;
    line-height: 1.7;
}

.info-section li {
    margin-bottom: 0.5rem;
}

/* Dark mode for info sections */
@media (prefers-color-scheme: dark) {
    .info-section {
        border-bottom-color: #4a5568;
    }
    
    .info-section h3 {
        color: #e2e8f0 !important;
    }
    
    .info-section p {
        color: #cbd5e0 !important;
    }
    
    .info-section ul {
        color: #cbd5e0 !important;
    }
}

.disclaimer {
    background: #fff3cd;
    padding: 1rem;
    font-size: 0.85rem;
    color: #856404;
    line-height: 1.5;
    margin-top: 1rem;
    border-radius: 6px;
}

/* Dark mode for disclaimer */
@media (prefers-color-scheme: dark) {
    .disclaimer {
        background: #744210;
        color: #fef3c7 !important;
    }
}

.error-msg, .error-container {
    background: #f8d7da;
    color: #721c24;
    padding: 1rem;
    border-radius: 6px;
    font-size: 0.9rem;
}

/* Dark mode for errors */
@media (prefers-color-scheme: dark) {
    .error-msg, .error-container {
        background: #742a2a;
        color: #feb2b2 !important;
    }
}

.model-stats {
    display: grid;
    grid-template-columns: repeat(4, 1fr);
    gap: 1rem;
    margin: 2rem 0;
    padding: 0 0.5rem;
}

.stat-card {
    background: #f8f9fa;
    padding: 1.25rem;
    border-radius: 6px;
    text-align: center;
    border: 1px solid #e9ecef;
}

.stat-value {
    font-size: 1.75rem;
    font-weight: 600;
    color: #2a5298;
    display: block;
}

.stat-label {
    font-size: 0.85rem;
    color: #6c757d;
    margin-top: 0.25rem;
}

/* Dark mode for stats */
@media (prefers-color-scheme: dark) {
    .stat-card {
        background: #2d3748;
        border-color: #4a5568;
    }
    
    .stat-value {
        color: #4299e1 !important;
    }
    
    .stat-label {
        color: #cbd5e0 !important;
    }
}

.footer-note {
    background: #f8f9fa;
    padding: 1.5rem;
    border-radius: 8px;
    margin-top: 2rem;
    text-align: center;
    border: 1px solid #e9ecef;
}

.footer-note h4 {
    font-size: 0.9rem;
    font-weight: 600;
    color: #212529;
    margin: 0 0 0.5rem 0;
}

.footer-note p {
    font-size: 0.85rem;
    color: #6c757d;
    line-height: 1.6;
    margin: 0 0 1rem 0;
}

.tech-stack {
    font-size: 0.8rem;
    color: #868e96;
    margin-top: 1rem;
}

/* Dark mode for footer */
@media (prefers-color-scheme: dark) {
    .footer-note {
        background: #2d3748;
        border-color: #4a5568;
    }
    
    .footer-note h4 {
        color: #e2e8f0 !important;
    }
    
    .footer-note p {
        color: #cbd5e0 !important;
    }
    
    .tech-stack {
        color: #a0aec0 !important;
    }
}

button {
    font-weight: 500 !important;
    border-radius: 6px !important;
}

.primary {
    background: linear-gradient(135deg, #1e3c72 0%, #2a5298 100%) !important;
    border: none !important;
}

footer {
    display: none !important;
}

/* Mobile Responsiveness */
@media (max-width: 768px) {
    .header h1 {
        font-size: 1.25rem;
    }
    
    .header p {
        font-size: 0.85rem;
    }
    
    .classes-grid {
        grid-template-columns: repeat(2, 1fr);
        gap: 0.75rem;
    }
    
    .model-stats {
        grid-template-columns: repeat(2, 1fr);
        gap: 0.75rem;
    }
    
    .result-header {
        grid-template-columns: 1fr;
        gap: 1rem;
    }
    
    .value {
        font-size: 1.25rem;
    }
    
    .stat-value {
        font-size: 1.5rem;
    }
    
    .gradio-container {
        padding: 0.5rem;
    }
    
    .header {
        padding: 1.5rem 1rem;
    }
}

@media (max-width: 480px) {
    .classes-grid {
        grid-template-columns: 1fr;
    }
    
    .model-stats {
        grid-template-columns: 1fr;
    }
    
    .header h1 {
        font-size: 1.1rem;
    }
    
    .header p {
        font-size: 0.8rem;
    }
}
"""

# Build interface
with gr.Blocks(css=custom_css, theme=gr.themes.Soft(), title="Poultry Disease Classifier") as demo:
    
    # Header
    gr.HTML("""
    <div class='header'>
        <h1>Poultry Disease Classification System</h1>
        <p>AI-powered diagnostic screening using EfficientNetV2 deep learning architecture</p>
    </div>
    """)
    
    # Disease classes
    gr.HTML("""
    <div class='classes-grid'>
        <div class='class-card'>
            <h4>Coccidiosis</h4>
            <p>Parasitic</p>
        </div>
        <div class='class-card'>
            <h4>Healthy</h4>
            <p>No disease</p>
        </div>
        <div class='class-card'>
            <h4>Newcastle</h4>
            <p>Viral</p>
        </div>
        <div class='class-card'>
            <h4>Salmonella</h4>
            <p>Bacterial</p>
        </div>
    </div>
    """)
    
    # Main interface
    with gr.Row():
        with gr.Column(scale=1):
            input_image = gr.Image(
                type="pil",
                label="Upload Image",
                height=400
            )
            
            with gr.Row():
                clear_btn = gr.ClearButton(
                    components=[input_image],
                    value="Clear",
                    size="lg"
                )
                predict_btn = gr.Button(
                    "Analyze",
                    variant="primary",
                    size="lg",
                    elem_classes="primary"
                )
            
            gr.HTML("""
            <div class='image-tips'>
                <h4>Image Guidelines</h4>
                <ul>
                    <li>Use clear, well-lit images</li>
                    <li>Ensure bird is clearly visible</li>
                    <li>Avoid blurry or dark photos</li>
                    <li>Close-up shots preferred</li>
                </ul>
            </div>
            """)
        
        with gr.Column(scale=1):
            output_label = gr.Label(
                label="Analysis Results - Classification Confidence",
                num_top_classes=4
            )
            output_text = gr.HTML()
    
    # Model information
    gr.HTML("""
    <div class='model-stats'>
        <div class='stat-card'>
            <span class='stat-value'>97%</span>
            <span class='stat-label'>Accuracy</span>
        </div>
        <div class='stat-card'>
            <span class='stat-value'>5.9M</span>
            <span class='stat-label'>Parameters</span>
        </div>
        <div class='stat-card'>
            <span class='stat-value'>360px</span>
            <span class='stat-label'>Input Size</span>
        </div>
        <div class='stat-card'>
            <span class='stat-value'>4</span>
            <span class='stat-label'>Classes</span>
        </div>
    </div>
    
    <div class='footer-note'>
        <h4>Clinical Notice</h4>
        <p>This diagnostic tool is intended for preliminary screening and educational purposes only. 
        Results should not replace professional veterinary examination and diagnosis. 
        Always consult a licensed veterinarian for accurate diagnosis and treatment recommendations.</p>
        <div class='tech-stack'>
        TensorFlow • Keras • EfficientNetV2-B0 • Gradio | Updated February 2026
        </div>
    </div>
    """)
    
    # Connect prediction
    predict_btn.click(
        fn=predict_disease,
        inputs=input_image,
        outputs=[output_label, output_text]
    )

if __name__ == "__main__":
    demo.launch(
        server_name="0.0.0.0",
        server_port=7860,
        share=True,
        show_error=True
    )