Update app.py

app.py

@@ -16,11 +16,6 @@ taxo_df['species'] = taxo_df['species'].str.replace('_', ' ')
 # Available taxonomic levels
 taxonomic_levels = ['species', 'genus', 'family', 'order', 'class']
 
-# Function to map predicted class index to class name at the selected taxonomic level
-def get_class_name(predicted_class, taxonomic_level):
-    unique_labels = sorted(taxo_df[taxonomic_level].unique())
-    return unique_labels[predicted_class]
-
 # Function to aggregate predictions to a higher taxonomic level
 def aggregate_predictions(predicted_probs, taxonomic_level, class_names):
     unique_labels = sorted(taxo_df[taxonomic_level].unique())
@@ -47,8 +42,8 @@ def load_and_preprocess_image(image, target_size=(224, 224)):
     img_array = preprocess_input(img_array)
     return img_array
 
-# Function to make predictions
-def make_prediction(image, taxonomic_level):
+# Function to make predictions when taxonomic level is specified
+def make_prediction_with_taxonomic_level(image, taxonomic_level):
     # Preprocess the image
     img_array = load_and_preprocess_image(image)
     
@@ -83,13 +78,12 @@ def make_prediction(image, taxonomic_level):
         
         if taxonomic_level == "species":
             # Display common names only at species level and make it italic
-            common_name = taxo_df[taxo_df[taxonomic_level] == class_name]['common_name'].values[0]
+            common_name = taxo_df[taxo_level == class_name]['common_name'].values[0]
             confidence_percentage = aggregated_predictions[0][i] * 100
             output_text += f"<div style='display: flex; justify-content: space-between;'>" \
                            f"<span style='font-style: italic;'>{class_name}</span>&nbsp;(<span>{common_name}</span>)" \
                            f"<span style='margin-left: auto;'>{confidence_percentage:.2f}%</span></div>"
         else:
-            # No common names at higher taxonomic levels
             confidence_percentage = aggregated_predictions[0][i] * 100
             output_text += f"<div style='display: flex; justify-content: space-between;'>" \
                            f"<span>{class_name}</span>" \
@@ -97,15 +91,72 @@ def make_prediction(image, taxonomic_level):
     
     return output_text
 
+# Function to make predictions with automatic taxonomic resolution
+def make_prediction_auto(image):
+    # Preprocess the image
+    img_array = load_and_preprocess_image(image)
+    
+    # Get the class names from the 'species' column
+    class_names = sorted(taxo_df['species'].unique())  
+    
+    # Make a prediction
+    prediction = model.predict(img_array)
+    
+    # Start with species-level predictions
+    taxonomic_level = 'species'
+    aggregated_predictions, aggregated_class_labels = aggregate_predictions(prediction, taxonomic_level, class_names)
+    
+    # Check confidence and move to higher taxonomic levels if necessary
+    predicted_class_index = np.argmax(aggregated_predictions)
+    confidence = aggregated_predictions[0][predicted_class_index]
+    
+    while confidence < 0.80 and taxonomic_levels.index(taxonomic_level) < len(taxonomic_levels) - 1:
+        # Move to the next higher taxonomic level
+        taxonomic_level = taxonomic_levels[taxonomic_levels.index(taxonomic_level) + 1]
+        aggregated_predictions, aggregated_class_labels = aggregate_predictions(prediction, taxonomic_level, class_names)
+        predicted_class_index = np.argmax(aggregated_predictions)
+        confidence = aggregated_predictions[0][predicted_class_index]
+
+    predicted_class_name = aggregated_class_labels[predicted_class_index]
+    
+    if taxonomic_level == "species":
+        predicted_common_name = taxo_df[taxo_df[taxonomic_level] == predicted_class_name]['common_name'].values[0]
+        output_text = f"<h1 style='font-weight: bold;'><span style='font-style: italic;'>{predicted_class_name}</span> ({predicted_common_name})</h1>"
+    else:
+        output_text = f"<h1 style='font-weight: bold;'>{predicted_class_name}</h1>"
+
+    # Return the final prediction text
+    return output_text
+
+# Gradio function to handle the flag logic
+def make_prediction(image, choose_resolution, taxonomic_level):
+    if choose_resolution == "Yes, I want to specify the taxonomic level":
+        return make_prediction_with_taxonomic_level(image, taxonomic_level)
+    else:
+        return make_prediction_auto(image)
+
 # Define the Gradio interface
 interface = gr.Interface(
-    fn=make_prediction,            # Function to be called for predictions
-    inputs=[gr.Image(type="pil"),   # Input type: Image (PIL format)
-            gr.Dropdown(choices=taxonomic_levels, label="Taxonomic level", value="species")],  # Use 'value' instead of 'default'
-    outputs="html",                # Output type: HTML for formatting
+    fn=make_prediction,            
+    inputs=[
+        gr.Image(type="pil"),  
+        gr.Radio(choices=["Yes, I want to specify the taxonomic level", "No, I will let the model decide"], 
+                 label="Do you want to choose the taxonomic resolution for predictions?", value="No, I will let the model decide"),
+        gr.Dropdown(choices=taxonomic_levels, label="Taxonomic level", value="species", interactive=True)],  
+    outputs="html",                
     title="Amazon arboreal species classification",
-    description="Upload an image and select the taxonomic level to classify the species."
+    description="Upload an image and select the taxonomic level or let the model decide the resolution."
 )
 
+# Add custom logic to disable the "Taxonomic level" dropdown when "No, I will let the model decide" is selected
+def update_taxonomic_level_interface(choose_resolution):
+    if choose_resolution == "No, I will let the model decide":
+        return gr.Dropdown.update(interactive=False)
+    else:
+        return gr.Dropdown.update(interactive=True)
+
+# Set up dynamic behavior for the interface
+interface.update(inputs=["Do you want to choose the taxonomic resolution for predictions?"], fn=update_taxonomic_level_interface)
+
 # Launch the Gradio interface
 interface.launch()