pandas.concat

app.py

@@ -9,7 +9,7 @@ import tensorflow as tf
 import tensorflow_hub as hub
 import io
 from sklearn.metrics.pairwise import cosine_similarity
-import tempfile  # Add this import
+import tempfile # Add this import
 import logging
 
 # Configure logging
@@ -36,12 +36,10 @@ def compute_textual_similarity(caption, statement):
     similarity_score = cosine_similarity([caption_embedding], [statement_embedding])[0][0]
     return similarity_score
 
-# List of statements for Image-Text Matching
 # Read statements from the external file 'statements.txt'
 with open('statements.txt', 'r') as file:
     statements = file.read().splitlines()
 
-
 # Function to compute ITM scores for the image-statement pair
 def compute_itm_score(image, statement):
     logging.info('Starting compute_itm_score')
@@ -66,12 +64,12 @@ def save_dataframe_to_csv(df):
     csv_buffer = io.StringIO()
     df.to_csv(csv_buffer, index=False)
     csv_string = csv_buffer.getvalue()
-    
+
     # Save the CSV string to a temporary file
     with tempfile.NamedTemporaryFile(mode="w+", delete=False, suffix=".csv") as temp_file:
         temp_file.write(csv_string)
-        temp_file_path = temp_file.name  # Get the file path
-    
+        temp_file_path = temp_file.name # Get the file path
+
     # Return the file path (no need to reopen the file with "rb" mode)
     return temp_file_path
 
@@ -86,38 +84,41 @@ def process_images_and_statements(image):
     weight_textual_similarity = 0.5
     weight_statement = 0.5
 
-    # Initialize an empty DataFrame with column names
-    results_df = pd.DataFrame(columns=['Statement', 'Generated Caption', 'Textual Similarity Score', 'ITM Score', 'Final Combined Score'])
+    # Initialize an empty list to store the results
+    results_list = []
 
     # Loop through each predefined statement
     for statement in statements:
         # Compute textual similarity between caption and statement
-        textual_similarity_score = (compute_textual_similarity(caption, statement) * 100)  # Multiply by 100
+        textual_similarity_score = (compute_textual_similarity(caption, statement) * 100) # Multiply by 100
 
         # Compute ITM score for the image-statement pair
-        itm_score_statement = (compute_itm_score(image, statement) * 100)  # Multiply by 100
+        itm_score_statement = (compute_itm_score(image, statement) * 100) # Multiply by 100
 
         # Combine the two scores using a weighted average
         final_score = ((weight_textual_similarity * textual_similarity_score) +
                        (weight_statement * itm_score_statement))
 
-        # Append the result to the DataFrame with formatted percentage values
-        results_df = results_df.append({
+        # Append the result to the results_list
+        results_list.append({
             'Statement': statement,
-            'Generated Caption': caption,  # Include the generated caption
-            'Textual Similarity Score': f"{textual_similarity_score:.2f}%",  # Format as percentage with two decimal places
-            'ITM Score': f"{itm_score_statement:.2f}%",  # Format as percentage with two decimal places
-            'Final Combined Score': f"{final_score:.2f}%"  # Format as percentage with two decimal places
-        }, ignore_index=True)
+            'Generated Caption': caption, # Include the generated caption
+            'Textual Similarity Score': f"{textual_similarity_score:.2f}%", # Format as percentage with two decimal places
+            'ITM Score': f"{itm_score_statement:.2f}%", # Format as percentage with two decimal places
+            'Final Combined Score': f"{final_score:.2f}%" # Format as percentage with two decimal places
+        })
+
+    # Convert the results_list to a DataFrame using pandas.concat
+    results_df = pd.concat([pd.DataFrame([result]) for result in results_list], ignore_index=True)
 
     logging.info('Finished process_images_and_statements')
 
     # Save results_df to a CSV file
     csv_results = save_dataframe_to_csv(results_df)
-    
+
     # Return both the DataFrame and the CSV data for the Gradio interface
-    return results_df, csv_results  # <--- Return results_df and csv_results
-    
+    return results_df, csv_results # <--- Return results_df and csv_results
+
 # Gradio interface
 image_input = gr.inputs.Image()
 output_df = gr.outputs.Dataframe(type="pandas", label="Results")
@@ -126,10 +127,10 @@ output_csv = gr.outputs.File(label="Download CSV")
 iface = gr.Interface(
     fn=process_images_and_statements,
     inputs=image_input,
-    outputs=[output_df, output_csv],  # Include both the DataFrame and CSV file outputs
+    outputs=[output_df, output_csv], # Include both the DataFrame and CSV file outputs
     title="Image Captioning and Image-Text Matching",
     theme='sudeepshouche/minimalist',
-    css=".output { flex-direction: column; } .output .outputs { width: 100%; }"  # Custom CSS
+    css=".output { flex-direction: column; } .output .outputs { width: 100%; }" # Custom CSS
 )
 
 iface.launch()