back to old code

app.py

@@ -5,8 +5,6 @@ from PIL import Image
 import google.generativeai as genai 
 import os
 import markdown2
-import cv2
-import matplotlib.cm as cm
 
 # Load the TensorFlow model
 model_path = 'model'
@@ -29,99 +27,64 @@ def get_disease_detail(disease_name):
         f"The diagnosis is {disease_name}, what are your recommendations?"
         )
     try:
-        response = genai.GenerativeModel("gemini-1.5-flash").generate_content(prompt)
+        response = genai.GenerativeModel("gemini-2.5-flash").generate_content(prompt)
         return markdown2.markdown(response.text.strip())
     except Exception as e:
         return f"Error: {e}"
 
-def generate_gradcam(model, img_array, layer_name, class_index):
-    grad_model = tf.keras.models.Model(
-        inputs=[model.inputs],
-        outputs=[model.get_layer(layer_name).output, model.output],
-    )
-
-    with tf.GradientTape() as tape:
-        conv_outputs, predictions = grad_model(img_array)
-        loss = predictions[:, class_index]
-
-    grads = tape.gradient(loss, conv_outputs)
-    pooled_grads = tf.reduce_mean(grads, axis=(0, 1, 2))
-
-    heatmap = tf.matmul(conv_outputs, tf.expand_dims(pooled_grads, axis=-1))[0]
-    heatmap = tf.reduce_mean(heatmap, axis=-1)
-
-    heatmap = np.maximum(heatmap, 0) / np.max(heatmap)
-    heatmap = cv2.resize(heatmap, (img_array.shape [1], img_array.shape [0]))
-    heatmap = np.uint8(255 * heatmap)
-    heatmap = cv2.applyColorMap(heatmap, cv2.COLORMAP_JET)
-
-    return heatmap
-
-def overlay_gradcam(img, heatmap, alpha=0.4):
-    img = img.astype('uint8')
-    resized_heatmap = cv2.resize(heatmap, (img.shape [1], img.shape [0]))
-    overlay = cv2.addWeighted(img, 1 - alpha, resized_heatmap, alpha, 0)
-    return overlay
-
-def predict_and_visualize(image):
+def predict_image(image):
     image_resized = image.resize((224, 224))
     image_array = np.array(image_resized).astype(np.float32) / 255.0
-    image_array_expanded = np.expand_dims(image_array, axis=0)
-    
-    predictions = model.signatures['serving_default'](tf.convert_to_tensor(image_array_expanded, dtype=tf.float32))['output_0']
+    image_array = np.expand_dims(image_array, axis=0)
+
+    predictions = model.signatures['serving_default'](tf.convert_to_tensor(image_array, dtype=tf.float32))['output_0']
     
+    # Highest prediction
     top_index = np.argmax(predictions.numpy(), axis=1)[0]
     top_label = labels[top_index]
     top_probability = predictions.numpy()[0][top_index]
-    prediction_dict = {top_label: top_probability}
-    
-    explanation_html = get_disease_detail(top_label)
-
-    last_conv_layer_name = 'top_activation' 
-    try:
-        heatmap = generate_gradcam(model, image_array_expanded, last_conv_layer_name, top_index)
-        gradcam_overlay = overlay_gradcam(np.array(image), heatmap) # Overlay ke gambar asli, bukan yg di-resize
-        gradcam_image = Image.fromarray(cv2.cvtColor(gradcam_overlay, cv2.COLOR_BGR2RGB))
-    except Exception as e:
-        print(f"Error saat membuat Grad-CAM: {e}")
-        gradcam_image = image
-
-    return gradcam_image, prediction_dict, explanation_html
 
-css = """.scrollable-html { height: 300px; overflow-y: auto; border: 1px solid #ccc; padding: 10px; box-sizing: border-box; }"""
-
-with gr.Blocks(css=css) as demo:
-    gr.Markdown("<h1>Eye Diseases Classifier</h1>")
-    gr.Markdown(
+    explanation = get_disease_detail(top_label)
+
+    return {top_label: top_probability}, explanation
+
+# Example images
+example_images = [
+    ["exp_eye_images/0_right_h.png"],
+    ["exp_eye_images/03fd50da928d_dr.png"],
+    ["exp_eye_images/108_right_h.png"],
+    ["exp_eye_images/1062_right_c.png"],
+    ["exp_eye_images/1084_right_c.png"],
+    ["exp_eye_images/image_1002_g.jpg"]
+]
+
+# Custom CSS for HTML height
+css = """
+.scrollable-html {
+    height: 206px;
+    overflow-y: auto;  
+    border: 1px solid #ccc;  
+    padding: 10px;  
+    box-sizing: border-box;
+}
+"""
+
+# Gradio Interface
+interface = gr.Interface(
+    fn=predict_image,
+    inputs=gr.Image(type="pil"),
+    outputs=[
+        gr.Label(num_top_classes=1, label="Prediction"), 
+        gr.Markdown(label="Explanation", elem_classes=["scrollable-html"])
+    ],
+    examples=example_images,
+    title="Eye Diseases Classifier",
+    description=(
         "Upload an image of an eye fundus, and the model will predict it.\n\n"
         "**Disclaimer:** This model is intended as a form of learning process in the field of health-related machine learning and was trained with a limited amount and variety of data with a total of about 4000 data, so the prediction results may not always be correct. There is still a lot of room for improvisation on this model in the future."
-    )
-    
-    with gr.Row():
-        with gr.Column(scale=1):
-            image_input = gr.Image(type="pil", label="Upload Eye Fundus Image")
-            detect_btn = gr.Button("Detect Disease", variant="primary")
-            gr.Examples(
-                examples=[
-                    ["exp_eye_images/0_right_h.png"],
-                    ["exp_eye_images/03fd50da928d_dr.png"],
-                    ["exp_eye_images/108_right_h.png"],
-                    ["exp_eye_images/1062_right_c.png"],
-                    ["exp_eye_images/1084_right_c.png"],
-                    ["exp_eye_images/image_1002_g.jpg"]
-                ],
-                inputs=image_input
-            )
-
-        with gr.Column(scale=1):
-            gradcam_output = gr.Image(label="Grad-CAM Visualization")
-            prediction_output = gr.Label(num_top_classes=1, label="Prediction")
-            explanation_output = gr.Markdown(label="Explanation", elem_classes=["scrollable-html"])
-
-    detect_btn.click(
-        fn=predict_and_visualize,
-        inputs=image_input,
-        outputs=[gradcam_output, prediction_output, explanation_output]
-    )
+    ),
+    allow_flagging="never",
+    css=css
+)
 
-demo.launch(share=True)
+interface.launch(share=True)