Update app.py

app.py

@@ -1,115 +1,58 @@
-import gradio as gr
-import numpy as np
+from dotenv import load_dotenv
+import os
+import google.generativeai as genai
 from PIL import Image
-from tensorflow.keras import models
-from tensorflow.keras.preprocessing.image import img_to_array
-import matplotlib.pyplot as plt
-import io
-import base64
-
-
-def inference(image, model_choice):
-    label_map = {'cassava-healthy': 0, 'cassava-not-healthy:bacteria blight': 1}
-    inverse_map = {v: k for k, v in label_map.items()}
-
-    image = Image.fromarray(image.astype('uint8'), 'RGB')
-    image = image.resize((64, 64))
-    image_arr = img_to_array(image)
-    image_arr /= 255
-    image_arr = image_arr[np.newaxis, :]
-
-    if model_choice == "Cassava Model 🍃":
-        model = models.load_model("cassava_model.keras")
-    else:
-        model = models.load_model("/content/maize_model.keras")
-    proba = model.predict(image_arr)
-    label = (proba > 0.5).squeeze().astype(int)
-
-    result = {
-        "label": inverse_map.get(int(label)),
-        "probability": float(proba.squeeze())
-    }
-
-    # Create visualization
-    fig, ax = plt.subplots(figsize=(8, 6))
-    ax.bar(['Healthy 🌿', 'Not Healthy 🍂'], [1 - result['probability'], result['probability']], color=['#2ecc71', '#e74c3c'])
-    ax.set_ylim(0, 1)
-    ax.set_ylabel('Probability')
-    ax.set_title('Plant Health Prediction 🔍', fontsize=16, fontweight='bold')
-    ax.spines['top'].set_visible(False)
-    ax.spines['right'].set_visible(False)
-    plt.tight_layout()
-
-    # Convert plot to image
-    buf = io.BytesIO()
-    plt.savefig(buf, format='png')
-    buf.seek(0)
-    plot_image = Image.open(buf)
-
-    return result["label"], f"{result['probability']:.2%} of illness(bacteria blight)", plot_image
+import gradio as gr
 
-# Custom CSS for styling
-custom_css = """
-#component-0 {
-    max-width: 730px;
-    margin: auto;
-    padding: 1.5rem;
-    border-radius: 10px;
-    background: linear-gradient(135deg, #f6d365 0%, #fda085 100%);
-    box-shadow: 0 10px 20px rgba(0,0,0,0.19), 0 6px 6px rgba(0,0,0,0.23);
-}
-#component-1 {
-    border-radius: 10px;
-    overflow: hidden;
-}
-#component-5 {
-    border-radius: 10px;
-    overflow: hidden;
-    box-shadow: 0 4px 6px rgba(0,0,0,0.1);
-}
-.label {
-    font-size: 18px !important;
-    color: #2c3e50;
-    font-weight: bold;
-}
-.output-class {
-    font-size: 24px !important;
-    color: #2980b9;
-    font-weight: bold;
-}
-.output-prob {
-    font-size: 20px !important;
-    color: #16a085;
-}
+# Load all the environment variables
+load_dotenv()
+genai.configure(api_key=os.getenv("GOOGLE_API_KEY"))
+
+# Function to load Google Gemini Pro Vision API and get a response
+def get_gemini_response(input_prompt, uploaded_image):
+    model = genai.GenerativeModel('gemini-1.5-flash')
+
+    # Convert the uploaded image to bytes
+    bytes_data = uploaded_image.read()
+    image_parts = [
+        {
+            "mime_type": uploaded_image.type,
+            "data": bytes_data
+        }
+    ]
+    
+    # Generate the content
+    response = model.generate_content([input_prompt, image_parts[0], ""])
+    return response.text
+
+# Input prompt for the model
+input_prompt = """
+    "You are an expert in computer vision and agriculture who can easily predict the disease of the plant. "
+    "Analyze the following image and provide 6 outputs in a structured table format: "
+    "1. Crop in the image, "
+    "2. Whether it is infected or healthy, "
+    "3. Type of disease (if any), "
+    "4. How confident out of 100% whether image is healthy or infected, "
+    "5. Reason for the disease such as whether it is happening due to fungus, bacteria, insect bite, poor nutrition, etc., "
+    "6. Precautions for it."
 """
 
-# Gradio interface
-with gr.Blocks(css=custom_css) as demo:
-    gr.Markdown("# 🌱 Crop Diseases Detector 🕵️‍♂️")
-    gr.Markdown("Upload an image of a cassava plant and let's check its health!")
-
-    with gr.Row():
-        input_image = gr.Image(type="numpy", label="📸 Upload or Capture Image")
-        output_image = gr.Image(type="pil", label="🖼️ Health Prediction Visualization")
-
-    model_choice = gr.Dropdown(["Cassava Model 🍃"], label="🤖 Select Model", value="Cassava Model 🍃")
-
-    with gr.Row():
-        detect_btn = gr.Button("🔍 Detect Plant Health", variant="primary")
-
-    output_label = gr.Textbox(label="🏷️ Diagnosis")
-    output_confidence = gr.Textbox(label="📊 Confidence")
-
-    detect_btn.click(
-        inference,
-        inputs=[input_image, model_choice],
-        outputs=[output_label, output_confidence, output_image]
-    )
-
-    gr.Markdown("## How to use:")
-    gr.Markdown("1. 📤 Upload an image or 📸 take a picture of a cassava plant")
-    gr.Markdown("2. 🤖 Select the model you want to use")
-    gr.Markdown("3. 🔍 Click 'Detect Plant Health' to get the results")
-    gr.Markdown("4. 📊 View the diagnosis, confidence score, and health prediction chart")
-
-demo.launch(debug=True)
+# Define the Gradio interface
+def predict_crop_health(uploaded_image):
+    if uploaded_image is None:
+        return "No image uploaded."
+    
+    response = get_gemini_response(input_prompt, uploaded_image)
+    return response
+
+# Create a Gradio interface
+iface = gr.Interface(
+    fn=predict_crop_health,
+    inputs=gr.Image(type="file", label="Upload Crop Image"),
+    outputs="text",
+    title="Gemini Crop Disease Detection App",
+    description="Upload an image of a crop to predict its health and identify any diseases."
+)
+
+# Launch the Gradio app
+iface.launch()