punimtag/test_deepface_gui.py

#!/usr/bin/env python3
"""
DeepFace GUI Test Application

GUI version of test_deepface_only.py that shows face comparison results
with left panel for reference faces and right panel for comparison faces with confidence scores.
"""

import os
import sys
import time
import tkinter as tk
from tkinter import ttk, messagebox, filedialog
from pathlib import Path
from typing import List, Dict, Tuple, Optional
import numpy as np
from PIL import Image, ImageTk

# Suppress TensorFlow warnings and CUDA errors
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
import warnings
warnings.filterwarnings('ignore')

# DeepFace library
from deepface import DeepFace

# Face recognition library
import face_recognition

# Supported image formats
SUPPORTED_FORMATS = {'.jpg', '.jpeg', '.png', '.bmp', '.tiff', '.tif'}


class FaceComparisonGUI:
    """GUI application for DeepFace face comparison testing"""
    
    def __init__(self):
        self.root = tk.Tk()
        self.root.title("Face Comparison Test - DeepFace vs face_recognition")
        self.root.geometry("2000x1000")
        self.root.minsize(1200, 800)
        
        # Data storage
        self.deepface_faces = []        # DeepFace faces from all images
        self.facerec_faces = []         # face_recognition faces from all images
        self.deepface_similarities = []  # DeepFace similarity results
        self.facerec_similarities = []   # face_recognition similarity results
        self.processing_times = {}       # Timing information for each photo
        
        # GUI components
        self.setup_gui()
        
    def setup_gui(self):
        """Set up the GUI layout"""
        # Main frame
        main_frame = ttk.Frame(self.root, padding="10")
        main_frame.grid(row=0, column=0, sticky=(tk.W, tk.E, tk.N, tk.S))
        
        # Configure grid weights
        self.root.columnconfigure(0, weight=1)
        self.root.rowconfigure(0, weight=1)
        main_frame.columnconfigure(0, weight=1)
        main_frame.rowconfigure(2, weight=1)  # Make the content area expandable
        
        # Title
        title_label = ttk.Label(main_frame, text="Face Comparison Test - DeepFace vs face_recognition", 
                               font=("Arial", 16, "bold"))
        title_label.grid(row=0, column=0, columnspan=3, pady=(0, 10))
        
        # Control panel
        control_frame = ttk.Frame(main_frame)
        control_frame.grid(row=1, column=0, columnspan=3, sticky=(tk.W, tk.E), pady=(0, 5))
        
        # Folder selection
        ttk.Label(control_frame, text="Test Folder:").grid(row=0, column=0, padx=(0, 5))
        self.folder_var = tk.StringVar(value="demo_photos/testdeepface/")
        folder_entry = ttk.Entry(control_frame, textvariable=self.folder_var, width=40)
        folder_entry.grid(row=0, column=1, padx=(0, 5))
        
        browse_btn = ttk.Button(control_frame, text="Browse", command=self.browse_folder)
        browse_btn.grid(row=0, column=2, padx=(0, 10))
        
        # Reference image selection
        ttk.Label(control_frame, text="Reference Image:").grid(row=0, column=3, padx=(10, 5))
        self.reference_var = tk.StringVar(value="2019-11-22_0011.JPG")
        reference_entry = ttk.Entry(control_frame, textvariable=self.reference_var, width=20)
        reference_entry.grid(row=0, column=4, padx=(0, 5))
        
        # Face detector selection
        ttk.Label(control_frame, text="Detector:").grid(row=0, column=5, padx=(10, 5))
        self.detector_var = tk.StringVar(value="retinaface")
        detector_combo = ttk.Combobox(control_frame, textvariable=self.detector_var, 
                                     values=["retinaface", "mtcnn", "opencv", "ssd"], 
                                     state="readonly", width=10)
        detector_combo.grid(row=0, column=6, padx=(0, 5))
        
        # Similarity threshold
        ttk.Label(control_frame, text="Threshold:").grid(row=0, column=7, padx=(10, 5))
        self.threshold_var = tk.StringVar(value="60")
        threshold_entry = ttk.Entry(control_frame, textvariable=self.threshold_var, width=8)
        threshold_entry.grid(row=0, column=8, padx=(0, 5))
        
        # Process button
        process_btn = ttk.Button(control_frame, text="Process Images", 
                                command=self.process_images, style="Accent.TButton")
        process_btn.grid(row=0, column=9, padx=(10, 0))
        
        # Progress bar
        self.progress_var = tk.DoubleVar()
        self.progress_bar = ttk.Progressbar(control_frame, variable=self.progress_var, 
                                           maximum=100, length=200)
        self.progress_bar.grid(row=1, column=0, columnspan=10, sticky=(tk.W, tk.E), pady=(5, 0))
        
        # Status label
        self.status_var = tk.StringVar(value="Ready to process images")
        status_label = ttk.Label(control_frame, textvariable=self.status_var)
        status_label.grid(row=2, column=0, columnspan=10, pady=(5, 0))
        
        # Main content area with three panels
        content_frame = ttk.Frame(main_frame)
        content_frame.grid(row=2, column=0, columnspan=3, sticky=(tk.W, tk.E, tk.N, tk.S), pady=(10, 0))
        content_frame.columnconfigure(0, weight=1)
        content_frame.columnconfigure(1, weight=1)
        content_frame.columnconfigure(2, weight=1)
        content_frame.rowconfigure(0, weight=1)
        
        # Left panel - DeepFace results
        left_frame = ttk.LabelFrame(content_frame, text="DeepFace Results", padding="5")
        left_frame.grid(row=0, column=0, sticky=(tk.W, tk.E, tk.N, tk.S), padx=(0, 5))
        left_frame.columnconfigure(0, weight=1)
        left_frame.rowconfigure(0, weight=1)
        
        # Left panel scrollable area
        left_canvas = tk.Canvas(left_frame, bg="white")
        left_scrollbar = ttk.Scrollbar(left_frame, orient="vertical", command=left_canvas.yview)
        self.left_scrollable_frame = ttk.Frame(left_canvas)
        
        self.left_scrollable_frame.bind(
            "<Configure>",
            lambda e: left_canvas.configure(scrollregion=left_canvas.bbox("all"))
        )
        
        left_canvas.create_window((0, 0), window=self.left_scrollable_frame, anchor="nw")
        left_canvas.configure(yscrollcommand=left_scrollbar.set)
        
        left_canvas.grid(row=0, column=0, sticky=(tk.W, tk.E, tk.N, tk.S))
        left_scrollbar.grid(row=0, column=1, sticky=(tk.N, tk.S))
        
        # Middle panel - face_recognition results
        middle_frame = ttk.LabelFrame(content_frame, text="face_recognition Results", padding="5")
        middle_frame.grid(row=0, column=1, sticky=(tk.W, tk.E, tk.N, tk.S), padx=(5, 5))
        middle_frame.columnconfigure(0, weight=1)
        middle_frame.rowconfigure(0, weight=1)
        
        # Right panel - Comparison Results
        right_frame = ttk.LabelFrame(content_frame, text="Comparison Results", padding="5")
        right_frame.grid(row=0, column=2, sticky=(tk.W, tk.E, tk.N, tk.S), padx=(5, 0))
        right_frame.columnconfigure(0, weight=1)
        right_frame.rowconfigure(0, weight=1)
        
        # Middle panel scrollable area
        middle_canvas = tk.Canvas(middle_frame, bg="white")
        middle_scrollbar = ttk.Scrollbar(middle_frame, orient="vertical", command=middle_canvas.yview)
        self.middle_scrollable_frame = ttk.Frame(middle_canvas)
        
        self.middle_scrollable_frame.bind(
            "<Configure>",
            lambda e: middle_canvas.configure(scrollregion=middle_canvas.bbox("all"))
        )
        
        middle_canvas.create_window((0, 0), window=self.middle_scrollable_frame, anchor="nw")
        middle_canvas.configure(yscrollcommand=middle_scrollbar.set)
        
        middle_canvas.grid(row=0, column=0, sticky=(tk.W, tk.E, tk.N, tk.S))
        middle_scrollbar.grid(row=0, column=1, sticky=(tk.N, tk.S))
        
        # Right panel scrollable area
        right_canvas = tk.Canvas(right_frame, bg="white")
        right_scrollbar = ttk.Scrollbar(right_frame, orient="vertical", command=right_canvas.yview)
        self.right_scrollable_frame = ttk.Frame(right_canvas)
        
        self.right_scrollable_frame.bind(
            "<Configure>",
            lambda e: right_canvas.configure(scrollregion=right_canvas.bbox("all"))
        )
        
        right_canvas.create_window((0, 0), window=self.right_scrollable_frame, anchor="nw")
        right_canvas.configure(yscrollcommand=right_scrollbar.set)
        
        right_canvas.grid(row=0, column=0, sticky=(tk.W, tk.E, tk.N, tk.S))
        right_scrollbar.grid(row=0, column=1, sticky=(tk.N, tk.S))
        
        # Bind mousewheel to all canvases
        def _on_mousewheel(event):
            left_canvas.yview_scroll(int(-1*(event.delta/120)), "units")
            middle_canvas.yview_scroll(int(-1*(event.delta/120)), "units")
            right_canvas.yview_scroll(int(-1*(event.delta/120)), "units")
        
        left_canvas.bind("<MouseWheel>", _on_mousewheel)
        middle_canvas.bind("<MouseWheel>", _on_mousewheel)
        right_canvas.bind("<MouseWheel>", _on_mousewheel)
        
    def browse_folder(self):
        """Browse for folder containing test images"""
        folder = filedialog.askdirectory(initialdir="demo_photos/")
        if folder:
            self.folder_var.set(folder)
    
    def update_status(self, message: str):
        """Update status message"""
        self.status_var.set(message)
        self.root.update_idletasks()
    
    def update_progress(self, value: float):
        """Update progress bar"""
        self.progress_var.set(value)
        self.root.update_idletasks()
    
    def get_image_files(self, folder_path: str) -> List[str]:
        """Get all supported image files from folder"""
        folder = Path(folder_path)
        if not folder.exists():
            raise FileNotFoundError(f"Folder not found: {folder_path}")
        
        image_files = []
        for file_path in folder.rglob("*"):
            if file_path.is_file() and file_path.suffix.lower() in SUPPORTED_FORMATS:
                image_files.append(str(file_path))
        
        return sorted(image_files)
    
    def process_with_deepface(self, image_path: str, detector: str = "retinaface") -> Dict:
        """Process image with DeepFace library"""
        try:
            # Use DeepFace.represent() to get proper face detection with regions
            # Using selected detector for face detection
            results = DeepFace.represent(
                img_path=image_path,
                model_name='ArcFace',  # Best accuracy model
                detector_backend=detector,  # User-selected detector
                enforce_detection=False,  # Don't fail if no faces
                align=True  # Face alignment for better accuracy
            )
            
            if not results:
                print(f"No faces found in {Path(image_path).name}")
                return {'faces': [], 'encodings': []}
            
            print(f"Found {len(results)} faces in {Path(image_path).name}")
            
            # Convert to our format
            faces = []
            encodings = []
            
            for i, result in enumerate(results):
                try:
                    # Extract face region info from DeepFace result
                    # DeepFace uses 'facial_area' instead of 'region'
                    facial_area = result.get('facial_area', {})
                    face_confidence = result.get('face_confidence', 0.0)
                    
                    # Create face data with proper bounding box
                    face_data = {
                        'image_path': image_path,
                        'face_id': f"df_{Path(image_path).stem}_{i}",
                        'location': (facial_area.get('y', 0), facial_area.get('x', 0) + facial_area.get('w', 0), 
                                   facial_area.get('y', 0) + facial_area.get('h', 0), facial_area.get('x', 0)),
                        'bbox': facial_area,
                        'encoding': np.array(result['embedding']),
                        'confidence': face_confidence
                    }
                    faces.append(face_data)
                    encodings.append(np.array(result['embedding']))
                    
                    print(f"Face {i}: facial_area={facial_area}, confidence={face_confidence:.2f}, embedding shape={np.array(result['embedding']).shape}")
                    
                except Exception as e:
                    print(f"Error processing face {i}: {e}")
                    continue
            
            return {
                'faces': faces,
                'encodings': encodings
            }
            
        except Exception as e:
            print(f"DeepFace error on {image_path}: {e}")
            return {'faces': [], 'encodings': []}
    
    def process_with_face_recognition(self, image_path: str) -> Dict:
        """Process image with face_recognition library"""
        try:
            # Load image
            image = face_recognition.load_image_file(image_path)
            
            # Find face locations
            face_locations = face_recognition.face_locations(image, model="hog")  # Use HOG model for speed
            
            if not face_locations:
                print(f"No faces found in {Path(image_path).name} (face_recognition)")
                return {'faces': [], 'encodings': []}
            
            print(f"Found {len(face_locations)} faces in {Path(image_path).name} (face_recognition)")
            
            # Get face encodings
            face_encodings = face_recognition.face_encodings(image, face_locations)
            
            # Convert to our format
            faces = []
            encodings = []
            
            for i, (face_location, face_encoding) in enumerate(zip(face_locations, face_encodings)):
                try:
                    # face_recognition returns (top, right, bottom, left)
                    top, right, bottom, left = face_location
                    
                    # Create face data with proper bounding box
                    face_data = {
                        'image_path': image_path,
                        'face_id': f"fr_{Path(image_path).stem}_{i}",
                        'location': face_location,
                        'bbox': {'x': left, 'y': top, 'w': right - left, 'h': bottom - top},
                        'encoding': np.array(face_encoding),
                        'confidence': 1.0  # face_recognition doesn't provide confidence scores
                    }
                    faces.append(face_data)
                    encodings.append(np.array(face_encoding))
                    
                    print(f"Face {i}: location={face_location}, encoding shape={np.array(face_encoding).shape}")
                    
                except Exception as e:
                    print(f"Error processing face {i}: {e}")
                    continue
            
            return {
                'faces': faces,
                'encodings': encodings
            }
            
        except Exception as e:
            print(f"face_recognition error on {image_path}: {e}")
            return {'faces': [], 'encodings': []}
    
    def extract_face_thumbnail(self, face_data: Dict, size: Tuple[int, int] = (150, 150)) -> ImageTk.PhotoImage:
        """Extract face thumbnail from image"""
        try:
            # Load original image
            image = Image.open(face_data['image_path'])
            
            # Extract face region
            bbox = face_data['bbox']
            left = bbox.get('x', 0)
            top = bbox.get('y', 0)
            right = left + bbox.get('w', 0)
            bottom = top + bbox.get('h', 0)
            
            # Add padding
            padding = 20
            left = max(0, left - padding)
            top = max(0, top - padding)
            right = min(image.width, right + padding)
            bottom = min(image.height, bottom + padding)
            
            # Crop face
            face_crop = image.crop((left, top, right, bottom))
            
            # FORCE resize to exact size (don't use thumbnail which maintains aspect ratio)
            face_crop = face_crop.resize(size, Image.Resampling.LANCZOS)
            
            print(f"DEBUG: Created thumbnail of size {face_crop.size} for {face_data['face_id']}")
            
            # Convert to PhotoImage
            return ImageTk.PhotoImage(face_crop)
            
        except Exception as e:
            print(f"Error extracting thumbnail for {face_data['face_id']}: {e}")
            # Return a placeholder image
            placeholder = Image.new('RGB', size, color='lightgray')
            return ImageTk.PhotoImage(placeholder)
    
    def calculate_face_similarity(self, encoding1: np.ndarray, encoding2: np.ndarray) -> float:
        """Calculate similarity between two face encodings using cosine similarity"""
        try:
            # Ensure encodings are numpy arrays
            enc1 = np.array(encoding1).flatten()
            enc2 = np.array(encoding2).flatten()
            
            # Check if encodings have the same length
            if len(enc1) != len(enc2):
                print(f"Warning: Encoding length mismatch: {len(enc1)} vs {len(enc2)}")
                return 0.0
            
            # Normalize encodings
            enc1_norm = enc1 / (np.linalg.norm(enc1) + 1e-8)  # Add small epsilon to avoid division by zero
            enc2_norm = enc2 / (np.linalg.norm(enc2) + 1e-8)
            
            # Calculate cosine similarity
            cosine_sim = np.dot(enc1_norm, enc2_norm)
            
            # Clamp cosine similarity to valid range [-1, 1]
            cosine_sim = np.clip(cosine_sim, -1.0, 1.0)
            
            # Convert to confidence percentage (0-100)
            # For face recognition, we typically want values between 0-100%
            # where higher values mean more similar faces
            confidence = max(0, min(100, (cosine_sim + 1) * 50))  # Scale from [-1,1] to [0,100]
            
            return confidence
            
        except Exception as e:
            print(f"Error calculating similarity: {e}")
            return 0.0
    
    def process_images(self):
        """Process all images and perform face comparison"""
        try:
            # Clear previous results
            self.deepface_faces = []
            self.facerec_faces = []
            self.deepface_similarities = []
            self.facerec_similarities = []
            self.processing_times = {}
            
            # Clear GUI panels
            for widget in self.left_scrollable_frame.winfo_children():
                widget.destroy()
            for widget in self.middle_scrollable_frame.winfo_children():
                widget.destroy()
            for widget in self.right_scrollable_frame.winfo_children():
                widget.destroy()
            
            folder_path = self.folder_var.get()
            threshold = float(self.threshold_var.get())
            
            if not folder_path:
                messagebox.showerror("Error", "Please specify folder path")
                return
            
            self.update_status("Getting image files...")
            self.update_progress(10)
            
            # Get all image files
            image_files = self.get_image_files(folder_path)
            if not image_files:
                messagebox.showerror("Error", "No image files found in the specified folder")
                return
            
            # Get selected detector
            detector = self.detector_var.get()
            
            self.update_status(f"Processing all images with both DeepFace and face_recognition...")
            self.update_progress(20)
            
            # Process all images with both libraries
            for i, image_path in enumerate(image_files):
                filename = Path(image_path).name
                self.update_status(f"Processing {filename}...")
                progress = 20 + (i / len(image_files)) * 50
                self.update_progress(progress)
                
                # Process with DeepFace
                start_time = time.time()
                deepface_result = self.process_with_deepface(image_path, detector)
                deepface_time = time.time() - start_time
                
                # Process with face_recognition
                start_time = time.time()
                facerec_result = self.process_with_face_recognition(image_path)
                facerec_time = time.time() - start_time
                
                # Store timing information
                self.processing_times[filename] = {
                    'deepface_time': deepface_time,
                    'facerec_time': facerec_time,
                    'total_time': deepface_time + facerec_time
                }
                
                # Store results
                self.deepface_faces.extend(deepface_result['faces'])
                self.facerec_faces.extend(facerec_result['faces'])
                
                print(f"Processed {filename}: DeepFace={deepface_time:.2f}s, face_recognition={facerec_time:.2f}s")
            
            if not self.deepface_faces and not self.facerec_faces:
                messagebox.showwarning("Warning", "No faces found in any images")
                return
            
            self.update_status("Calculating face similarities...")
            self.update_progress(75)
            
            # Calculate similarities for DeepFace
            for i, face1 in enumerate(self.deepface_faces):
                similarities = []
                for j, face2 in enumerate(self.deepface_faces):
                    if i != j:  # Don't compare face with itself
                        confidence = self.calculate_face_similarity(
                            face1['encoding'], face2['encoding']
                        )
                        if confidence >= threshold:  # Only include faces above threshold
                            similarities.append({
                                'face': face2,
                                'confidence': confidence
                            })
                
                # Sort by confidence (highest first)
                similarities.sort(key=lambda x: x['confidence'], reverse=True)
                self.deepface_similarities.append({
                    'face': face1,
                    'similarities': similarities
                })
            
            # Calculate similarities for face_recognition
            for i, face1 in enumerate(self.facerec_faces):
                similarities = []
                for j, face2 in enumerate(self.facerec_faces):
                    if i != j:  # Don't compare face with itself
                        confidence = self.calculate_face_similarity(
                            face1['encoding'], face2['encoding']
                        )
                        if confidence >= threshold:  # Only include faces above threshold
                            similarities.append({
                                'face': face2,
                                'confidence': confidence
                            })
                
                # Sort by confidence (highest first)
                similarities.sort(key=lambda x: x['confidence'], reverse=True)
                self.facerec_similarities.append({
                    'face': face1,
                    'similarities': similarities
                })
            
            self.update_status("Displaying results...")
            self.update_progress(95)
            
            # Display results in GUI
            self.display_results()
            
            total_deepface_faces = len(self.deepface_faces)
            total_facerec_faces = len(self.facerec_faces)
            avg_deepface_time = sum(t['deepface_time'] for t in self.processing_times.values()) / len(self.processing_times)
            avg_facerec_time = sum(t['facerec_time'] for t in self.processing_times.values()) / len(self.processing_times)
            
            self.update_status(f"Complete! DeepFace: {total_deepface_faces} faces ({avg_deepface_time:.2f}s avg), face_recognition: {total_facerec_faces} faces ({avg_facerec_time:.2f}s avg)")
            self.update_progress(100)
            
        except Exception as e:
            messagebox.showerror("Error", f"Processing failed: {str(e)}")
            self.update_status("Error occurred during processing")
            print(f"Error: {e}")
            import traceback
            traceback.print_exc()
    
    def display_results(self):
        """Display the face comparison results in the GUI panels"""
        # Display DeepFace results in left panel
        self.display_library_results(self.deepface_similarities, self.left_scrollable_frame, "DeepFace")
        
        # Display face_recognition results in middle panel
        self.display_library_results(self.facerec_similarities, self.middle_scrollable_frame, "face_recognition")
        
        # Display timing comparison in right panel
        self.display_timing_comparison()
    
    def display_library_results(self, similarities_list: List[Dict], parent_frame, library_name: str):
        """Display results for a specific library"""
        for i, result in enumerate(similarities_list):
            face = result['face']
            
            # Create frame for this face
            face_frame = ttk.Frame(parent_frame)
            face_frame.grid(row=i, column=0, sticky=(tk.W, tk.E), pady=5, padx=5)
            
            # Face thumbnail
            thumbnail = self.extract_face_thumbnail(face, size=(80, 80))
            thumbnail_label = ttk.Label(face_frame, image=thumbnail)
            thumbnail_label.image = thumbnail  # Keep a reference
            thumbnail_label.grid(row=0, column=0, padx=5, pady=5)
            
            # Face info
            info_frame = ttk.Frame(face_frame)
            info_frame.grid(row=0, column=1, sticky=(tk.W, tk.E), padx=5)
            
            ttk.Label(info_frame, text=f"Face {i+1}", font=("Arial", 10, "bold")).grid(row=0, column=0, sticky=tk.W, pady=1)
            ttk.Label(info_frame, text=f"ID: {face['face_id']}", font=("Arial", 8)).grid(row=1, column=0, sticky=tk.W, pady=1)
            ttk.Label(info_frame, text=f"Image: {Path(face['image_path']).name}", font=("Arial", 8)).grid(row=2, column=0, sticky=tk.W, pady=1)
            
            # Show number of similar faces
            similar_count = len(result['similarities'])
            ttk.Label(info_frame, text=f"Similar: {similar_count}", font=("Arial", 8, "bold")).grid(row=3, column=0, sticky=tk.W, pady=1)
    
    def display_timing_comparison(self):
        """Display timing comparison between libraries"""
        if not self.processing_times:
            return
        
        # Create summary frame
        summary_frame = ttk.LabelFrame(self.right_scrollable_frame, text="Processing Times Summary")
        summary_frame.grid(row=0, column=0, sticky=(tk.W, tk.E), pady=5, padx=5)
        
        # Calculate averages
        total_deepface_time = sum(t['deepface_time'] for t in self.processing_times.values())
        total_facerec_time = sum(t['facerec_time'] for t in self.processing_times.values())
        avg_deepface_time = total_deepface_time / len(self.processing_times)
        avg_facerec_time = total_facerec_time / len(self.processing_times)
        
        # Summary statistics
        ttk.Label(summary_frame, text=f"Total Images: {len(self.processing_times)}", font=("Arial", 10, "bold")).grid(row=0, column=0, sticky=tk.W, pady=2)
        ttk.Label(summary_frame, text=f"DeepFace Avg: {avg_deepface_time:.2f}s", font=("Arial", 9)).grid(row=1, column=0, sticky=tk.W, pady=1)
        ttk.Label(summary_frame, text=f"face_recognition Avg: {avg_facerec_time:.2f}s", font=("Arial", 9)).grid(row=2, column=0, sticky=tk.W, pady=1)
        
        speed_ratio = avg_deepface_time / avg_facerec_time if avg_facerec_time > 0 else 0
        if speed_ratio > 1:
            faster_lib = "face_recognition"
            speed_text = f"{speed_ratio:.1f}x faster"
        else:
            faster_lib = "DeepFace"
            speed_text = f"{1/speed_ratio:.1f}x faster"
        
        ttk.Label(summary_frame, text=f"{faster_lib} is {speed_text}", font=("Arial", 9, "bold"), foreground="green").grid(row=3, column=0, sticky=tk.W, pady=2)
        
        # Individual photo timings
        timing_frame = ttk.LabelFrame(self.right_scrollable_frame, text="Per-Photo Timing")
        timing_frame.grid(row=1, column=0, sticky=(tk.W, tk.E), pady=5, padx=5)
        
        row = 0
        for filename, times in sorted(self.processing_times.items()):
            ttk.Label(timing_frame, text=f"{filename[:20]}...", font=("Arial", 8)).grid(row=row, column=0, sticky=tk.W, pady=1)
            ttk.Label(timing_frame, text=f"DF: {times['deepface_time']:.2f}s", font=("Arial", 8)).grid(row=row, column=1, sticky=tk.W, pady=1, padx=(5,0))
            ttk.Label(timing_frame, text=f"FR: {times['facerec_time']:.2f}s", font=("Arial", 8)).grid(row=row, column=2, sticky=tk.W, pady=1, padx=(5,0))
            row += 1
    
    def display_comparison_faces(self, ref_index: int, similarities: List[Dict]):
        """Display comparison faces for a specific reference face"""
        # Create frame for this reference face's comparisons
        comp_frame = ttk.LabelFrame(self.right_scrollable_frame, 
                                   text=f"Matches for Reference Face {ref_index + 1}")
        comp_frame.grid(row=ref_index, column=0, sticky=(tk.W, tk.E), pady=10, padx=10)
        
        # Display top matches (limit to avoid too much clutter)
        max_matches = min(8, len(similarities))
        
        for i in range(max_matches):
            sim_data = similarities[i]
            face = sim_data['face']
            confidence = sim_data['confidence']
            
            # Create frame for this comparison face
            face_frame = ttk.Frame(comp_frame)
            face_frame.grid(row=i, column=0, sticky=(tk.W, tk.E), pady=5, padx=10)
            
            # Face thumbnail
            thumbnail = self.extract_face_thumbnail(face, size=(120, 120))
            thumbnail_label = ttk.Label(face_frame, image=thumbnail)
            thumbnail_label.image = thumbnail  # Keep a reference
            thumbnail_label.grid(row=0, column=0, padx=10, pady=5)
            
            # Face info with confidence
            info_frame = ttk.Frame(face_frame)
            info_frame.grid(row=0, column=1, sticky=(tk.W, tk.E), padx=10)
            
            # Confidence with color coding
            confidence_text = f"{confidence:.1f}%"
            if confidence >= 80:
                confidence_color = "green"
            elif confidence >= 60:
                confidence_color = "orange"
            else:
                confidence_color = "red"
            
            ttk.Label(info_frame, text=confidence_text, 
                     font=("Arial", 14, "bold"), foreground=confidence_color).grid(row=0, column=0, sticky=tk.W, pady=2)
            ttk.Label(info_frame, text=f"ID: {face['face_id']}", font=("Arial", 10)).grid(row=1, column=0, sticky=tk.W, pady=2)
            ttk.Label(info_frame, text=f"Image: {Path(face['image_path']).name}", font=("Arial", 10)).grid(row=2, column=0, sticky=tk.W, pady=2)
    
    def run(self):
        """Start the GUI application"""
        self.root.mainloop()


def main():
    """Main entry point"""
    # Check dependencies
    try:
        from deepface import DeepFace
    except ImportError as e:
        print(f"Error: Missing required dependency: {e}")
        print("Please install with: pip install deepface")
        sys.exit(1)
    
    try:
        import face_recognition
    except ImportError as e:
        print(f"Error: Missing required dependency: {e}")
        print("Please install with: pip install face_recognition")
        sys.exit(1)
    
    # Suppress TensorFlow warnings and errors
    import os
    os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'  # Suppress TensorFlow warnings
    import warnings
    warnings.filterwarnings('ignore')
    
    try:
        # Create and run GUI
        app = FaceComparisonGUI()
        app.run()
    except Exception as e:
        print(f"GUI Error: {e}")
        import traceback
        traceback.print_exc()
        sys.exit(1)


if __name__ == "__main__":
    main()