app.py

from ultralytics import YOLO
import gradio as gr
import numpy as np
from PIL import Image

# Load YOLO model
model = YOLO("best.pt")

# Define what class names indicate a human
HUMAN_CLASSES = ["person", "human"]

# Predict Function
def predict(image):
    image_status = "✅ Real Image"
    disease_status = "✅ No Disease Detected"
    details = ""
    audio_html = "alert-1.mp3"  # for autoplay sound

    try:
        results = model(image)
        boxes = results[0].boxes
        class_names = model.names

        found_human = False
        found_disease = False

        if len(boxes) > 0:
            for box in boxes:
                class_id = int(box.cls[0])
                class_name = class_names[class_id].lower()
                confidence = float(box.conf[0])
                details += f"🔎 {class_name} - {confidence*100:.2f}%\n"

                if class_name in HUMAN_CLASSES:
                    found_human = True
                else:
                    found_disease = True

        if found_human:
            disease_status = "🚨 Human Detected (Invalid Input)"
            audio_html = """<audio autoplay><source src="file/disease_alert.mp3" type="audio/mpeg"></audio>"""
        elif found_disease:
            disease_status = "🚨 Disease Detected"
            audio_html = """<audio autoplay><source src="file/disease_alert.mp3" type="audio/mpeg"></audio>"""
        else:
            disease_status = "✅ No Disease Detected"

    except Exception as e:
        image_status = "❌ Error"
        disease_status = "⚠️ Detection Failed"
        details = str(e)

    return image_status, disease_status, details, audio_html


# Gradio Interface
with gr.Blocks(title="🚨 AgroScan - Human Alert + Disease Detection") as demo:
    gr.Markdown("""
    <h1 style="text-align:center; color:#2e7d32;">🌿 AgroScan</h1>
    <p style="text-align:center;">
        Upload an image of a plant leaf. If a human or disease is detected, an alert will be triggered with sound 🚨.
    </p>
    """)

    with gr.Row():
        with gr.Column(scale=1):
            image_input = gr.Image(type="pil", label="📷 Upload Image")
            submit_btn = gr.Button("🔍 Analyze")
        with gr.Column(scale=1):
            result1 = gr.Textbox(label="🖼️ Image Status")
            result2 = gr.Textbox(label="🌱 Disease Status")
            result3 = gr.Textbox(label="📋 Detection Details", lines=5)
            audio_out = gr.HTML(label="🚨 Auto Alert") # autoplay sound (no controls)

    submit_btn.click(fn=predict, inputs=image_input,
                     outputs=[result1, result2, result3, audio_out])

demo.launch()





# from ultralytics import YOLO
# import gradio as gr
# import numpy as np
# from PIL import Image

# # Load YOLO model
# model = YOLO("best.pt")

# # Define human classes
# HUMAN_CLASSES = ["person", "human"]

# # Prediction function
# def predict(image):
#     image_status = "✅ Real Image"
#     disease_status = "✅ No Disease Detected"
#     details = ""
#     alert_sound = "alert-1.mp3"  # Will hold sound file path

#     try:
#         results = model(image)
#         boxes = results[0].boxes
#         class_names = model.names

#         found_human = False
#         found_disease = False

#         if len(boxes) > 0:
#             for box in boxes:
#                 class_id = int(box.cls[0])
#                 class_name = class_names[class_id].lower()
#                 confidence = float(box.conf[0])
#                 details += f"🔎 {class_name} - {confidence*100:.2f}%\n"

#                 if class_name in HUMAN_CLASSES:
#                     found_human = True
#                 else:
#                     found_disease = True

#         if found_human:
#             disease_status = "🚨 Human Detected (Invalid Input)"
#             alert_sound = "disease_alert.mp3"
#         elif found_disease:
#             disease_status = "🚨 Disease Detected"
#             alert_sound = "disease_alert.mp3"
#         else:
#             disease_status = "✅ No Disease Detected"

#     except Exception as e:
#         image_status = "❌ Error"
#         disease_status = "⚠️ Detection Failed"
#         details = str(e)

#     return image_status, disease_status, details, alert_sound

# # Gradio Interface
# with gr.Blocks(title="🚨 AgroScan - Human Alert + Disease Detection") as demo:
#     gr.Markdown("""
#     <h1 style="text-align:center; color:#2e7d32;">🌿 AgroScan</h1>
#     <p style="text-align:center;">
#         Upload an image of a plant leaf. If a human or disease is detected, an alert will be triggered with sound 🚨.
#     </p>
#     """)

#     with gr.Row():
#         with gr.Column(scale=1):
#             image_input = gr.Image(type="pil", label="📷 Upload Image")
#             submit_btn = gr.Button("🔍 Analyze")
#         with gr.Column(scale=1):
#             result1 = gr.Textbox(label="🖼️ Image Status")
#             result2 = gr.Textbox(label="🌱 Disease Status")
#             result3 = gr.Textbox(label="📋 Detection Details", lines=5)
#             audio_out = gr.Audio(label="🚨 Alert Sound", autoplay=True)

#     submit_btn.click(fn=predict, inputs=image_input,
#                      outputs=[result1, result2, result3, audio_out])

# demo.launch()




# from ultralytics import YOLO
# import gradio as gr
# import numpy as np
# from PIL import Image

# # Load YOLO model
# model = YOLO("best.pt")

# # Define what class names indicate a human
# HUMAN_CLASSES = ["person", "human"]

# # Predict Function
# def predict(image):
#     image_status = "✅ Real Image"
#     disease_status = "✅ No Disease Detected"
#     details = ""
#     alert_sound = None  # path to MP3 file

#     try:
#         results = model(image)
#         boxes = results[0].boxes
#         class_names = model.names

#         found_human = False
#         found_disease = False

#         if len(boxes) > 0:
#             for box in boxes:
#                 class_id = int(box.cls[0])
#                 class_name = class_names[class_id].lower()
#                 confidence = float(box.conf[0])
#                 details += f"🔎 {class_name} - {confidence*100:.2f}%\n"

#                 if class_name in HUMAN_CLASSES:
#                     found_human = True
#                 else:
#                     found_disease = True

#         if found_human:
#             disease_status = "🚨 Human Detected (Invalid Input)"
#             alert_sound = "disease_alert.mp3"
#         elif found_disease:
#             disease_status = "🚨 Disease Detected"
#             alert_sound = "disease_alert.mp3"
#         else:
#             disease_status = "✅ No Disease Detected"

#     except Exception as e:
#         image_status = "❌ Error"
#         disease_status = "⚠️ Detection Failed"
#         details = str(e)

#     return image_status, disease_status, details, alert_sound

# # Gradio Interface
# with gr.Blocks(title="🚨 AgroScan - Human Alert + Disease Detection") as demo:
#     gr.Markdown("""
#     <h1 style="text-align:center; color:#2e7d32;">🌿 AgroScan</h1>
#     <p style="text-align:center;">
#         Upload an image of a plant leaf. If a human or disease is detected, an alert will be triggered with sound 🚨.
#     </p>
#     """)

#     with gr.Row():
#         with gr.Column(scale=1):
#             image_input = gr.Image(type="pil", label="📷 Upload Image")
#             submit_btn = gr.Button("🔍 Analyze")
#         with gr.Column(scale=1):
#             result1 = gr.Textbox(label="🖼️ Image Status")
#             result2 = gr.Textbox(label="🌱 Disease Status")
#             result3 = gr.Textbox(label="📋 Detection Details", lines=5)
#             audio_out = gr.Audio(label="🚨 Alert Sound", autoplay=True)

#     submit_btn.click(fn=predict, inputs=image_input,
#                      outputs=[result1, result2, result3, audio_out])

# demo.launch()





# from ultralytics import YOLO
# import gradio as gr
# import numpy as np
# from PIL import Image

# # Load the trained YOLOv8 model
# model = YOLO("best.pt")

# # Class name that represents "human" in your YOLO model
# HUMAN_CLASS_NAMES = ["person", "human"]

# # Prediction function with sound logic
# def predict(image):
#     image_status = "✅ Real Image"
#     disease_status = "✅ No Disease Detected"
#     disease_details = ""
#     play_alert = None  # MP3 to be played

#     try:
#         results = model(image)
#         boxes = results[0].boxes
#         names = model.names

#         found_human = False
#         if len(boxes) > 0:
#             for box in boxes:
#                 class_id = int(box.cls[0])
#                 class_name = names[class_id]
#                 confidence = float(box.conf[0])
#                 disease_details += f"🔍 {class_name} - {confidence*100:.2f}%\n"

#                 if class_name.lower() in HUMAN_CLASS_NAMES:
#                     found_human = True

#         if found_human:
#             disease_status = "🚨 Disease Detected (Human Image)"
#             play_alert = "disease_alert.mp3"  # make sure this file exists
#         elif len(boxes) == 0:
#             disease_status = "✅ No Disease Detected"
#         else:
#             disease_status = "🚨 Disease Detected"
#             play_alert = "disease_alert.mp3"

#     except Exception as e:
#         disease_status = "❌ Error during detection"
#         disease_details = str(e)

#     return image_status, disease_status, disease_details, play_alert

# # Gradio UI
# with gr.Blocks(title="AgroScan - Human Alert + Disease Detection") as demo:
#     gr.Markdown("""
#     <div style="text-align: center;">
#         <h1 style="color: #2e7d32;">🌿 AgroScan: Plant Disease Detector + Human Alert</h1>
#         <p style="font-size: 16px; color: #555;">
#             Upload an image of a leaf or a human. If human is detected, an alert will be triggered.
#         </p>
#     </div>
#     """)

#     with gr.Row():
#         with gr.Column(scale=1):
#             image_input = gr.Image(type="pil", label="📷 Upload Image")
#             submit_btn = gr.Button("🔍 Analyze")
#         with gr.Column(scale=1):
#             image_result = gr.Textbox(label="🖼️ Image Authenticity", interactive=False)
#             disease_result = gr.Textbox(label="🌱 Disease Status", interactive=False)
#             detail = gr.Textbox(label="📋 Details", lines=5, interactive=False)
#             sound_output = gr.Audio(label="🚨 Alert Sound", autoplay=True, interactive=False)

#     submit_btn.click(fn=predict, inputs=image_input,
#                      outputs=[image_result, disease_result, detail, sound_output])

# demo.launch()




# from ultralytics import YOLO
# import gradio as gr
# import numpy as np
# from PIL import Image

# # Load the trained YOLOv8 model
# model = YOLO("best.pt")  # Make sure 'best.pt' is in the same directory

# # Check if image is real (simple check using NumPy array and format)
# def is_real_image(image):
#     try:
#         if image is None:
#             return False
#         arr = np.array(image)
#         if arr.ndim == 3 and arr.shape[2] in [3, 4]:  # RGB or RGBA
#             return True
#         return False
#     except Exception as e:
#         print(f"Image check error: {e}")
#         return False

# # Prediction function
# def predict(image):
#     image_status = "❌ Fake Image"
#     disease_status = "⚠️ Unknown"
#     disease_details = ""

#     # Step 1: Image authenticity check
#     if is_real_image(image):
#         image_status = "✅ Real Image"
#     else:
#         return image_status, "⚠️ Cannot detect disease in a fake image.", ""

#     # Step 2: YOLOv8 disease detection
#     try:
#         results = model(image)
#         boxes = results[0].boxes
#         names = model.names

#         if len(boxes) == 0:
#             disease_status = "✅ No Disease Detected"
#         else:
#             disease_status = "🚨 Disease Detected"
#             for box in boxes:
#                 class_id = int(box.cls[0])
#                 class_name = names[class_id]
#                 confidence = float(box.conf[0])
#                 disease_details += f"🔬 {class_name} - Confidence: {confidence*100:.2f}%\n"
#     except Exception as e:
#         disease_status = "❌ Error during detection"
#         disease_details = str(e)

#     return image_status, disease_status, disease_details

# # Gradio UI
# with gr.Blocks(title="AgroScan - Plant Disease & Image Verifier") as demo:
#     gr.Markdown("""
#     <div style="text-align: center;">
#         <h1 style="color: #2e7d32;">🌿 AgroScan: Plant Disease Detector + Image Validator</h1>
#         <p style="font-size: 16px; color: #555;">
#             Upload a plant leaf image 🍃 to check if it's real or fake and detect any diseases using YOLOv8.
#         </p>
#     </div>
#     """)

#     with gr.Row():
#         with gr.Column(scale=1):
#             image_input = gr.Image(type="pil", label="📷 Upload Leaf Image")
#             submit_btn = gr.Button("🔍 Analyze")
#         with gr.Column(scale=1):
#             image_result = gr.Textbox(label="🖼️ Image Authenticity", interactive=False)
#             disease_result = gr.Textbox(label="🌱 Disease Status", interactive=False)
#             detail = gr.Textbox(label="📋 Detailed Diagnosis", lines=5, interactive=False)

#     with gr.Accordion("ℹ️ What is AgroScan?", open=False):
#         gr.Markdown("""
#         AgroScan is an AI-powered detector for identifying plant diseases from leaf images. 
#         It also verifies whether the image is genuine (real leaf image) or fake to avoid misdiagnosis. 
#         Designed to support farmers, agronomists, and researchers for better crop care.
#         """)

#     submit_btn.click(fn=predict, inputs=image_input,
#                      outputs=[image_result, disease_result, detail])

# demo.launch()





# from ultralytics import YOLO
# import gradio as gr
# from PIL import Image

# # Load YOLO model
# model = YOLO("best.pt")  # Make sure 'best.pt' is in your directory

# # Function to check if image is real
# def is_real_image(img):
#     try:
#         return isinstance(img, Image.Image) and img.mode in ['RGB', 'RGBA'] and img.size[0] > 50 and img.size[1] > 50
#     except:
#         return False

# # Main prediction function with error handling
# def predict(image):
#     try:
#         if image is None:
#             return "❌ No image provided", "⚠️ Cannot check disease", "Upload a valid image first."

#         # Step 1: Check image authenticity
#         if is_real_image(image):
#             image_status = "✅ Real Image"
#         else:
#             return "❌ Fake Image", "⚠️ Cannot check disease", "Image doesn't appear real. Try another image."

#         # Step 2: Predict with YOLO model
#         results = model(image)
#         names = model.names
#         boxes = results[0].boxes

#         if len(boxes) == 0:
#             return image_status, "✅ No Disease Detected", "No signs of disease found."
        
#         diagnosis = ""
#         disease_found = False
#         for box in boxes:
#             conf = float(box.conf[0])
#             if conf > 0.5:
#                 cls_id = int(box.cls[0])
#                 disease_name = names[cls_id]
#                 diagnosis += f"🔬 {disease_name} - Confidence: {conf*100:.2f}%\n"
#                 disease_found = True
        
#         if disease_found:
#             return image_status, "🚨 Disease Detected", diagnosis
#         else:
#             return image_status, "✅ No Disease Detected", "No confident detections."

#     except Exception as e:
#         # Handle unexpected errors gracefully
#         return "❌ Error", "❌ Error", f"An error occurred: {str(e)}"

# # Gradio Interface
# with gr.Blocks(title="AgroScan - Plant Disease & Image Verifier") as demo:
#     gr.Markdown("## 🌿 AgroScan: Plant Disease Detector + Image Validator")

#     with gr.Row():
#         with gr.Column():
#             image_input = gr.Image(type="pil", label="📷 Upload Plant Leaf Image")
#             submit_btn = gr.Button("🔍 Analyze")
#         with gr.Column():
#             image_result = gr.Textbox(label="🖼️ Image Authenticity", interactive=False)
#             disease_result = gr.Textbox(label="🌱 Disease Status", interactive=False)
#             detail = gr.Textbox(label="📋 Detailed Diagnosis", lines=5, interactive=False)

#     submit_btn.click(fn=predict, inputs=image_input, outputs=[image_result, disease_result, detail])

# demo.launch()







# from ultralytics import YOLO
# import gradio as gr
# from PIL import Image
# import imghdr
# import os

# # Load YOLOv5 model
# model = YOLO("best.pt")  # Ensure your model file 'best.pt' is in the same directory

# # Check if image is real or fake (basic format check)
# def is_real_image(image):
#     try:
#         if image is None:
#             return False
#         format_check = imghdr.what(image.fp.name)
#         return format_check in ['jpeg', 'png']
#     except:
#         return False

# # Prediction function
# def predict(image):
#     image_status = "❌ Fake Image"
#     disease_status = "⚠️ Unknown"
#     disease_details = ""

#     # Step 1: Check if image is real
#     if is_real_image(image):
#         image_status = "✅ Real Image"
#     else:
#         return image_status, "⚠️ Cannot detect disease in a fake image.", ""

#     # Step 2: Run disease detection
#     results = model(image)
#     boxes = results[0].boxes
#     names = model.names

#     if len(boxes) == 0:
#         disease_status = "✅ No Disease Detected"
#     else:
#         disease_status = "🚨 Disease Detected"
#         for box in boxes:
#             class_id = int(box.cls[0])
#             class_name = names[class_id]
#             confidence = float(box.conf[0])
#             disease_details += f"🔬 {class_name} - Confidence: {confidence*100:.2f}%\n"

#     return image_status, disease_status, disease_details

# # UI using Gradio Blocks
# with gr.Blocks(title="AgroScan - Plant Disease & Image Verifier") as demo:
#     gr.Markdown("""
#     <div style="text-align: center;">
#         <h1 style="color: #2e7d32;">🌿 AgroScan: Plant Disease Detector + Image Validator</h1>
#         <p style="font-size: 16px; color: #555;">
#             Upload a plant leaf image 🍃 to check if it's real or fake and detect any diseases using YOLOv5.
#         </p>
#     </div>
#     """)

#     with gr.Row():
#         with gr.Column(scale=1):
#             image_input = gr.Image(type="pil", label="📷 Upload Leaf Image")
#             submit_btn = gr.Button("🔍 Analyze")
#         with gr.Column(scale=1):
#             image_result = gr.Textbox(label="🖼️ Image Authenticity", interactive=False)
#             disease_result = gr.Textbox(label="🌱 Disease Status", interactive=False)
#             detail = gr.Textbox(label="📋 Detailed Diagnosis", lines=5, interactive=False)

#     with gr.Accordion("ℹ️ What is AgroScan?", open=False):
#         gr.Markdown("""
#         AgroScan is an AI-powered detector for identifying plant diseases from leaf images. 
#         It also verifies whether the image is genuine (real leaf image) or fake to avoid misdiagnosis. 
#         Designed to support farmers, agronomists, and researchers for better crop care.
#         """)

#     submit_btn.click(fn=predict, inputs=image_input,
#                      outputs=[image_result, disease_result, detail])

# demo.launch()







# from ultralytics import YOLO
# import gradio as gr
# from PIL import Image

# # Load YOLOv5 model
# model = YOLO("best.pt")  # Make sure this file exists in your Hugging Face Space

# # Prediction function
# def predict(image):
#     results = model(image)
#     boxes = results[0].boxes
#     names = model.names
#     output = ""

#     if len(boxes) == 0:
#         output = "✅ Healthy: No disease detected!"
#     else:
#         for box in boxes:
#             class_id = int(box.cls[0])
#             class_name = names[class_id]
#             confidence = float(box.conf[0])
#             output += f"🚨 Detected: {class_name} ({confidence*100:.2f}%)\n"
#     return output

# # Gradio UI using Blocks
# with gr.Blocks(title="AgroScan - Plant Disease Detector") as demo:
#     gr.Markdown("""
#     <div style="text-align: center;">
#         <h1 style="color: #2e7d32;">🌿 AgroScan: Plant Disease Detector</h1>
#         <p style="font-size: 16px; color: #555;">
#             Upload a high-quality image of a plant leaf 🍃 and detect possible diseases using AI-powered YOLOv5.
#         </p>
#     </div>
#     """)
    
#     with gr.Row():
#         with gr.Column(scale=1):
#             image_input = gr.Image(type="pil", label="📷 Upload Leaf Image")
#             predict_button = gr.Button("🔍 Analyze Leaf", variant="primary")
#         with gr.Column(scale=1):
#             result_output = gr.Textbox(label="🩺 Diagnosis Result", lines=6)

#     with gr.Accordion("ℹ️ About AgroScan", open=False):
#         gr.Markdown("""
#         **AgroScan** is an intelligent plant disease detector built with **YOLOv5** and trained on custom agricultural datasets. 
#         It helps farmers, researchers, and agriculturists in early detection of plant diseases to ensure timely treatment and improved crop yield. 🌱
#         """)

#     predict_button.click(fn=predict, inputs=image_input, outputs=result_output)

# demo.launch()






# from ultralytics import YOLO
# import gradio as gr
# from PIL import Image
# import torch

# # Load YOLOv5 model
# model = YOLO("best.pt")  # assumes model is in the same directory

# # Prediction function
# def predict(image):
#     results = model(image)
#     boxes = results[0].boxes
#     names = model.names
#     output = ""

#     if len(boxes) == 0:
#         output = "✅ Healthy: No disease detected!"
#     else:
#         for box in boxes:
#             class_id = int(box.cls[0])
#             class_name = names[class_id]
#             confidence = float(box.conf[0])
#             output += f"🚨 Detected: {class_name} ({confidence*100:.2f}%)\n"

#     return output

# # Gradio UI
# gr.Interface(
#     fn=predict,
#     inputs=gr.Image(type="pil"),
#     outputs="text",
#     title="🌿 AgroScan: Plant Disease Detector",
#     description="Upload a plant leaf image to detect disease using a YOLOv5 model.",
# ).launch()