app.py

from flask import Flask, request, jsonify, render_template_string, send_from_directory, send_file
from flask_cors import CORS
import os
import importlib.util
import time
import zipfile
import shutil
import json
from datetime import datetime
from predict_task2 import Predictor
from id_mapping import mapping
from show_stitched import *
import cv2
import supervision as sv
from ultralytics import YOLO

app = Flask(__name__)
CORS(app)

# Load configuration
config_dir = os.path.abspath(os.path.dirname(__file__))
config_path = os.path.join(config_dir, 'PC_CONFIG.py')
spec = importlib.util.spec_from_file_location("PC_CONFIG", config_path)
PC_CONFIG = importlib.util.module_from_spec(spec)
spec.loader.exec_module(PC_CONFIG)

HOST = PC_CONFIG.HOST
PORT = 7860  # Changed from PC_CONFIG.IMAGE_REC_PORT to 7860
UPLOAD_FOLDER = os.path.join(PC_CONFIG.FILE_DIRECTORY, "image-rec", "images")
DATASET_FOLDER = os.path.join(PC_CONFIG.BASE_DIR, "yolo_dataset")
ANNOTATED_FOLDER = os.path.join(DATASET_FOLDER, "annotated_images")
LABELS_FOLDER = os.path.join(DATASET_FOLDER, "labels")
IMAGES_FOLDER = os.path.join(DATASET_FOLDER, "images")
CLASS_MAPPING_FILE = os.path.join(DATASET_FOLDER, "classes.json")

app.config['UPLOAD_FOLDER'] = UPLOAD_FOLDER

# Initialize predictor
predictor = Predictor()

# Ensure directories exist
os.makedirs(UPLOAD_FOLDER, exist_ok=True)
os.makedirs(DATASET_FOLDER, exist_ok=True)
os.makedirs(ANNOTATED_FOLDER, exist_ok=True)
os.makedirs(LABELS_FOLDER, exist_ok=True)
os.makedirs(IMAGES_FOLDER, exist_ok=True)

# Initialize class mapping file
if not os.path.exists(CLASS_MAPPING_FILE):
    # Create initial class mapping from id_mapping.py
    reverse_mapping = {str(v): k for k, v in mapping.items() if v not in [-999] and k is not None}
    with open(CLASS_MAPPING_FILE, 'w', encoding='utf-8') as f:
        json.dump(reverse_mapping, f, indent=2, ensure_ascii=False)

# Create YOLO to ID mapping compatibility function
def get_compatible_class_id(yolo_class_name):
    """Map YOLO class names to id_mapping values"""
    # Handle different naming conventions between YOLO model and id_mapping
    yolo_to_id_mapping = {
        # Numbers
       # "one": 11, "two": 12, "three": 13, "four": 14, "five": 15,
        #"six": 16, "seven": 17, "eight": 18, "nine": 19,
        # Letters
       # "A": 20, "B": 21, "C": 22, "D": 23, "E": 24, "F": 25,
       # "G": 26, "H": 27, "S": 28, "T": 29, "U": 30, "V": 31,
       # "W": 32, "X": 33, "Y": 34, "Z": 35,
        # Directions
        "up": 36, "down": 37, "right": 38, "left": 39,
        # Shapes
        "circle": 40, "Bullseye": -1, "bullseye": -1
    }
    
    return yolo_to_id_mapping.get(yolo_class_name, mapping.get(yolo_class_name, -999))

def load_class_mapping():
    """Load class mapping from JSON file"""
    try:
        with open(CLASS_MAPPING_FILE, 'r', encoding='utf-8') as f:
            return json.load(f)
    except:
        return {}

def save_class_mapping(class_mapping):
    """Save class mapping to JSON file"""
    with open(CLASS_MAPPING_FILE, 'w', encoding='utf-8') as f:
        json.dump(class_mapping, f, indent=2, ensure_ascii=False)

def save_obstacle_image(detection_result, obstacle_num):
    """Save obstacle image for display"""
    if not detection_result or not detection_result.get('marked_image_path'):
        return
    
    # Create obstacles display folder
    obstacles_folder = os.path.join(PC_CONFIG.BASE_DIR, "obstacles_display")
    os.makedirs(obstacles_folder, exist_ok=True)
    
    # Copy annotated image to obstacles display folder
    source_path = detection_result['marked_image_path']
    target_filename = f"obstacle_{obstacle_num}.jpg"
    target_path = os.path.join(obstacles_folder, target_filename)
    
    # Save detection info to JSON file for later retrieval
    info_filename = f"obstacle_{obstacle_num}_info.json"
    info_path = os.path.join(obstacles_folder, info_filename)
    
    try:
        if os.path.exists(source_path):
            shutil.copy2(source_path, target_path)
            
            # Save detection info
            detection_info = {
                'label': detection_result.get('label', 'Unknown'),
                'image_id': detection_result.get('image_id', 'N/A'),
                'confidence': detection_result.get('confidence', 0.0),
                'timestamp': time.time()
            }
            
            with open(info_path, 'w', encoding='utf-8') as f:
                json.dump(detection_info, f, indent=2, ensure_ascii=False)
            
            print(f"Saved obstacle {obstacle_num} image and info to {target_path}")
    except Exception as e:
        print(f"Error saving obstacle image: {e}")

def get_obstacle_detection_info(obstacle_num):
    """Get detection info for a specific obstacle"""
    obstacles_folder = os.path.join(PC_CONFIG.BASE_DIR, "obstacles_display")
    info_filename = f"obstacle_{obstacle_num}_info.json"
    info_path = os.path.join(obstacles_folder, info_filename)
    
    try:
        if os.path.exists(info_path):
            with open(info_path, 'r', encoding='utf-8') as f:
                return json.load(f)
    except Exception as e:
        print(f"Error loading obstacle info: {e}")
    
    return {}

def generate_yolo_annotation(results, detection_id, image_width, image_height, class_name):
    """Generate YOLO format annotation string"""
    if not results or not results[0].boxes or detection_id >= len(results[0].boxes):
        return ""
    
    # Get class mapping
    class_mapping = load_class_mapping()
    
    # Get class ID from mapping, if not found, add it
    class_id = None
    for id_str, name in class_mapping.items():
        if name == class_name:
            class_id = int(id_str)
            break
    
    if class_id is None:
        # Add new class to mapping
        max_id = max([int(k) for k in class_mapping.keys()]) if class_mapping else -1
        class_id = max_id + 1
        class_mapping[str(class_id)] = class_name
        save_class_mapping(class_mapping)
    
    # Get bounding box - handle tensor conversion properly
    try:
        box = results[0].boxes.xyxy[detection_id]
        if hasattr(box, 'cpu'):
            box = box.cpu()
        if hasattr(box, 'numpy'):
            box = box.numpy()
        x1, y1, x2, y2 = box.tolist() if hasattr(box, 'tolist') else box
        
        # Convert to YOLO format (normalized)
        x_center = ((x1 + x2) / 2) / image_width
        y_center = ((y1 + y2) / 2) / image_height
        width = (x2 - x1) / image_width
        height = (y2 - y1) / image_height
        
        confidence = results[0].boxes.conf[detection_id]
        if hasattr(confidence, 'item'):
            confidence = confidence.item()
        
        # YOLO format: class_id x_center y_center width height
        return f"{class_id} {x_center:.6f} {y_center:.6f} {width:.6f} {height:.6f}"
    except Exception as e:
        print(f"Error generating YOLO annotation: {e}")
        return ""

def save_annotated_image(image, results, detection_id, filename):
    """Save annotated image with bounding boxes"""
    if not results or not results[0].boxes:
        return None
    
    try:
        # Simple annotation without supervision library
        annotated_image = image.copy()
        
        # Get detection boxes and info
        boxes = results[0].boxes.xyxy
        if hasattr(boxes, 'cpu'):
            boxes = boxes.cpu().numpy()
        
        confidences = results[0].boxes.conf
        if hasattr(confidences, 'cpu'):
            confidences = confidences.cpu().numpy()
            
        class_ids = results[0].boxes.cls
        if hasattr(class_ids, 'cpu'):
            class_ids = class_ids.cpu().numpy().astype(int)
        
        # Draw bounding boxes
        for i, (box, conf, cls_id) in enumerate(zip(boxes, confidences, class_ids)):
            x1, y1, x2, y2 = map(int, box)
            class_name = results[0].names[cls_id]
            
            # Draw rectangle
            cv2.rectangle(annotated_image, (x1, y1), (x2, y2), (0, 255, 0), 2)
            
            # Draw label
            label = f"{class_name} {conf:.2f}"
            label_size = cv2.getTextSize(label, cv2.FONT_HERSHEY_SIMPLEX, 0.5, 1)[0]
            cv2.rectangle(annotated_image, (x1, y1 - label_size[1] - 10), 
                         (x1 + label_size[0], y1), (0, 255, 0), -1)
            cv2.putText(annotated_image, label, (x1, y1 - 5), 
                       cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 0), 1)
        
        # Save annotated image with timestamp to ensure proper ordering
        annotated_path = os.path.join(ANNOTATED_FOLDER, f"annotated_{filename}")
        cv2.imwrite(annotated_path, annotated_image)
        
        # Clean up old images if more than 4 exist
        cleanup_old_annotated_images()
        
        return annotated_path
        
    except Exception as e:
        print(f"Error saving annotated image: {e}")
        # Fallback: just copy the original image
        annotated_path = os.path.join(ANNOTATED_FOLDER, f"annotated_{filename}")
        cv2.imwrite(annotated_path, image)
        cleanup_old_annotated_images()
        return annotated_path

def cleanup_old_annotated_images():
    """Keep only the latest 4 annotated images"""
    try:
        if not os.path.exists(ANNOTATED_FOLDER):
            return
        
        # Get all annotated images
        annotated_files = []
        for filename in os.listdir(ANNOTATED_FOLDER):
            if filename.startswith('annotated_') and filename.lower().endswith(('.png', '.jpg', '.jpeg')):
                filepath = os.path.join(ANNOTATED_FOLDER, filename)
                mtime = os.path.getmtime(filepath)
                annotated_files.append((filename, filepath, mtime))
        
        # If we have more than 4 images, remove the oldest ones
        if len(annotated_files) > 4:
            # Sort by modification time (newest first)
            annotated_files.sort(key=lambda x: x[2], reverse=True)
            
            # Remove files beyond the first 4
            for filename, filepath, _ in annotated_files[4:]:
                try:
                    os.remove(filepath)
                    print(f"Removed old annotated image: {filename}")
                except Exception as e:
                    print(f"Error removing old image {filename}: {e}")
    
    except Exception as e:
        print(f"Error during cleanup: {e}")

def process_file(file_path, direction, task_type, filename):
    """Process uploaded file and generate predictions"""
    print("File received and saved successfully.")
    print(f"Direction received: {direction}")
    print(f"Task type received: {task_type}")

    startTime = datetime.now()
    
    # Load image and ensure it's 640x640
    image = cv2.imread(file_path)
    if image is None:
        return None
    
    # Resize image to 640x640 if needed (to match YOLO model input)
    if image.shape[0] != 640 or image.shape[1] != 640:
        image = cv2.resize(image, (640, 640))
        # Save the resized image back to ensure consistency
        cv2.imwrite(file_path, image)
    
    # Perform prediction
    class_name, results, detection_id = predictor.predict_id(file_path, task_type)
    
    # For TASK_2, apply priority-based selection AFTER getting all detections
    if task_type == "TASK_2" and results and results[0].boxes is not None and len(results[0].boxes) > 0:
        detections_list = []
        
        boxes = results[0].boxes
        for i in range(len(boxes)):
            detected_class = results[0].names[int(boxes.cls[i])]
            confidence = float(boxes.conf[i])
            yolo_class_id = int(boxes.cls[i])
            
            print(f"[APP.PY] Detection {i}: {detected_class} (confidence: {confidence:.2f}, class_id: {yolo_class_id})")
            
            # Only set Bullseye to lowest priority, all others have equal priority (0)
            # Check by class name to be model-agnostic
            if detected_class.lower() == 'bullseye':
                priority = -10  # Lowest priority for bullseye
            else:
                priority = 0  # Equal priority for all non-bullseye detections
            
            detections_list.append({
                'index': i,
                'class_name': detected_class,
                'confidence': confidence,
                'priority': priority,
                'yolo_class_id': yolo_class_id
            })
        
        if detections_list:
            # Sort by priority (descending), then by confidence (descending)
            detections_list.sort(key=lambda x: (x['priority'], x['confidence']), reverse=True)
            
            print(f"\n[APP.PY] Sorted detections:")
            for det in detections_list:
                print(f"  - {det['class_name']}: priority={det['priority']}, confidence={det['confidence']:.2f}")
            
            # Override with highest priority detection
            selected = detections_list[0]
            class_name = selected['class_name']
            detection_id = selected['index']
            print(f"\n[APP.PY] ✓ Final selection: {class_name} (priority: {selected['priority']}, confidence: {selected['confidence']:.2f})")
    
    # Use compatible mapping function
    class_id = str(get_compatible_class_id(class_name))
    
    detection_result = None
    if class_name and results and results[0].boxes is not None and len(results[0].boxes) > 0:
        # Generate filename
        timestamp = int(time.time())
        base_filename = f"{class_name}_{timestamp}"
        
        # Save original image to dataset
        image_filename = f"{base_filename}.jpg"
        dataset_image_path = os.path.join(IMAGES_FOLDER, image_filename)
        shutil.copy2(file_path, dataset_image_path)
        
        # Generate and save YOLO annotation
        h, w = image.shape[:2]
        yolo_annotation = generate_yolo_annotation(results, detection_id, w, h, class_name)
        txt_path = None
        if yolo_annotation:
            txt_filename = f"{base_filename}.txt"
            txt_path = os.path.join(LABELS_FOLDER, txt_filename)
            with open(txt_path, 'w') as f:
                f.write(yolo_annotation)
        
        # Save annotated image
        annotated_path = save_annotated_image(image, results, detection_id, image_filename)
        
        # Get bounding box and confidence for compatibility
        try:
            box = results[0].boxes.xyxy[detection_id]
            if hasattr(box, 'cpu'):
                box = box.cpu()
            if hasattr(box, 'numpy'):
                box = box.numpy()
            x1, y1, x2, y2 = box.tolist() if hasattr(box, 'tolist') else box
            
            confidence = results[0].boxes.conf[detection_id]
            if hasattr(confidence, 'item'):
                confidence = confidence.item()
            else:
                confidence = float(confidence)
        except Exception as e:
            print(f"Error extracting box/confidence: {e}")
            x1, y1, x2, y2 = 0, 0, 0, 0
            confidence = 0.0
        
        # Create detection result in compatible format
        detection_result = {
            "image_id": class_id,
            "label": class_name,
            "confidence": confidence,
            "bbox": [x1, y1, x2, y2],
            "original_image_path": dataset_image_path,
            "marked_image_path": annotated_path,
            "txt_file_path": txt_path
        }
    else:
        # Handle case when no detection found
        print("No valid detections found")
        # Still save the image for record keeping
        timestamp = int(time.time())
        base_filename = f"no_detection_{timestamp}"
        image_filename = f"{base_filename}.jpg"
        dataset_image_path = os.path.join(IMAGES_FOLDER, image_filename)
        shutil.copy2(file_path, dataset_image_path)
        
        detection_result = {
            "image_id": class_id,
            "label": class_name or "unknown",
            "confidence": 0.0,
            "bbox": [0, 0, 0, 0],
            "original_image_path": dataset_image_path,
            "marked_image_path": dataset_image_path,  # Use original as no annotation
            "txt_file_path": None
        }
    
    endTime = datetime.now()
    totalTime = (endTime - startTime).total_seconds()
    print(f"Predicted ID: {class_id}")
    print(f"Time taken for Predicting Image = {totalTime} s")
    
    return class_id, detection_result

# HTML template for the frontend
HTML_TEMPLATE = """
<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>YOLO Image Recognition System</title>
    <style>
        body {
            font-family: Arial, sans-serif;
            max-width: 1200px;
            margin: 0 auto;
            padding: 20px;
            background-color: #f5f5f5;
        }
        .header {
            text-align: center;
            margin-bottom: 30px;
        }
        .container {
            display: grid;
            grid-template-columns: 1fr 1fr;
            gap: 20px;
            margin-bottom: 30px;
        }
        .card {
            background-color: white;
            padding: 20px;
            border-radius: 10px;
            box-shadow: 0 2px 10px rgba(0,0,0,0.1);
        }
        .images-display {
            grid-column: span 2;
        }
        .images-grid {
            display: grid;
            grid-template-columns: repeat(4, 1fr);
            gap: 15px;
            margin-top: 15px;
        }
        .image-slot {
            background-color: #f8f9fa;
            border: 2px dashed #dee2e6;
            border-radius: 8px;
            padding: 10px;
            text-align: center;
            min-height: 200px;
            display: flex;
            flex-direction: column;
            justify-content: center;
            align-items: center;
        }
        .image-slot.has-image {
            border: 2px solid #007bff;
            border-style: solid;
            background-color: white;
        }
        .image-slot img {
            max-width: 100%;
            max-height: 150px;
            border-radius: 5px;
            box-shadow: 0 2px 5px rgba(0,0,0,0.1);
        }
        .image-label {
            margin-top: 8px;
            font-size: 14px;
            font-weight: bold;
            color: #007bff;
        }
        .image-timestamp {
            font-size: 12px;
            color: #666;
            margin-top: 5px;
        }
        .empty-slot {
            color: #6c757d;
            font-style: italic;
        }
        .status {
            padding: 10px;
            border-radius: 5px;
            margin-bottom: 15px;
            font-weight: bold;
        }
        .status.waiting {
            background-color: #fff3cd;
            color: #856404;
        }
        .status.updated {
            background-color: #d4edda;
            color: #155724;
        }
        .upload-section {
            text-align: center;
        }
        .upload-form {
            display: inline-block;
            text-align: left;
        }
        .form-group {
            margin-bottom: 15px;
        }
        .form-group label {
            display: block;
            margin-bottom: 5px;
            font-weight: bold;
        }
        .form-group input, .form-group select {
            width: 300px;
            padding: 8px;
            border: 1px solid #ddd;
            border-radius: 4px;
        }
        .btn {
            background-color: #007bff;
            color: white;
            border: none;
            padding: 12px 24px;
            border-radius: 5px;
            font-size: 16px;
            cursor: pointer;
            transition: background-color 0.3s;
        }
        .btn:hover {
            background-color: #0056b3;
        }
        .btn:disabled {
            background-color: #6c757d;
            cursor: not-allowed;
        }
        .dataset-info {
            grid-column: span 2;
        }
        .stats-grid {
            display: grid;
            grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
            gap: 15px;
            margin-top: 15px;
        }
        .stat-card {
            background-color: #f8f9fa;
            padding: 15px;
            border-radius: 5px;
            text-align: center;
        }
        .stat-number {
            font-size: 24px;
            font-weight: bold;
            color: #007bff;
        }
        .timestamp {
            color: #666;
            font-size: 12px;
            margin-top: 10px;
        }
        .class-mapping {
            max-height: 200px;
            overflow-y: auto;
            background-color: #f8f9fa;
            padding: 10px;
            border-radius: 5px;
            font-family: monospace;
            font-size: 12px;
        }
    </style>
</head>
<body>
    <div class="header">
        <h1>YOLO Image Recognition System</h1>
        <p>Real-time image processing with dataset management</p>
        <div style="margin-top: 15px;">
            <a href="/" style="margin: 0 10px; padding: 8px 16px; background-color: #007bff; color: white; text-decoration: none; border-radius: 4px;">Main Interface</a>
            <a href="/obstacles" style="margin: 0 10px; padding: 8px 16px; background-color: #28a745; color: white; text-decoration: none; border-radius: 4px;">Obstacles Display</a>
        </div>
    </div>

    <div class="container">
        <div class="card images-display">
            <h3>Recent Recognition Results (Latest 4 Images)</h3>
            <div id="status" class="status waiting">Waiting for recognition results...</div>
            <div class="images-grid" id="imagesGrid">
                <div class="image-slot" id="slot1">
                    <div class="empty-slot">Waiting for first image...</div>
                </div>
                <div class="image-slot" id="slot2">
                    <div class="empty-slot">Waiting for second image...</div>
                </div>
                <div class="image-slot" id="slot3">
                    <div class="empty-slot">Waiting for third image...</div>
                </div>
                <div class="image-slot" id="slot4">
                    <div class="empty-slot">Waiting for fourth image...</div>
                </div>
            </div>
        </div>

        <div class="card">
            <h3>Upload Image for Recognition</h3>
            <div class="upload-section">
                <form id="uploadForm" class="upload-form" enctype="multipart/form-data">
                    <div class="form-group">
                        <label for="file">Select Image:</label>
                        <input type="file" id="file" name="file" accept="image/*" required>
                    </div>
                    <div class="form-group">
                        <label for="direction">Direction:</label>
                        <select id="direction" name="direction" required>
                            <option value="north">North</option>
                            <option value="south">South</option>
                            <option value="east">East</option>
                            <option value="west">West</option>
                        </select>
                    </div>
                    <div class="form-group">
                        <label for="task_type">Task Type:</label>
                        <select id="task_type" name="task_type" required>
                            <option value="TASK_1">Task 1</option>
                            <option value="TASK_2" selected>Task 2</option>
                        </select>
                    </div>
                    <div class="form-group">
                        <label for="num_obstacles">Obstacle Number (Optional):</label>
                        <select id="num_obstacles" name="NUM_OBSTACLES">
                            <option value="0">None</option>
                            <option value="1">Obstacle 1</option>
                            <option value="2">Obstacle 2</option>
                            <option value="3">Obstacle 3</option>
                            <option value="4">Obstacle 4</option>
                            <option value="5">Obstacle 5</option>
                            <option value="6">Obstacle 6</option>
                            <option value="7">Obstacle 7</option>
                            <option value="8">Obstacle 8</option>
                        </select>
                    </div>
                    <button type="submit" class="btn">Upload and Predict</button>
                </form>
                <div id="uploadResult" style="margin-top: 15px;"></div>
            </div>
        </div>

        <div class="card dataset-info">
            <h3>Dataset Information</h3>
            <div class="stats-grid">
                <div class="stat-card">
                    <div class="stat-number" id="totalImages">0</div>
                    <div>Total Images</div>
                </div>
                <div class="stat-card">
                    <div class="stat-number" id="totalClasses">0</div>
                    <div>Total Classes</div>
                </div>
                <div class="stat-card">
                    <div class="stat-number" id="annotatedImages">0</div>
                    <div>Annotated Images</div>
                </div>
                <div class="stat-card">
                    <button id="downloadBtn" class="btn" onclick="downloadDataset()">
                        Download Dataset
                    </button>
                </div>
            </div>
            <div style="margin-top: 20px;">
                <h4>Class Mapping:</h4>
                <div id="classMapping" class="class-mapping">Loading...</div>
            </div>
            <div id="downloadInfo" style="margin-top: 10px; color: #666;"></div>
        </div>
    </div>

    <script>
        let imageHistory = [];
        
        // Check for latest results
        async function checkLatestImages() {
            try {
                const response = await fetch('/latest-images');
                const data = await response.json();
                
                if (data.success && data.images) {
                    updateImageDisplay(data.images);
                }
            } catch (error) {
                console.error('Failed to check latest images:', error);
            }
        }
        
        // Update image display with latest 4 images
        function updateImageDisplay(images) {
            const slots = ['slot1', 'slot2', 'slot3', 'slot4'];
            
            // Update image history
            imageHistory = images;
            
            // Update status
            const statusDiv = document.getElementById('status');
            if (images.length > 0) {
                statusDiv.className = 'status updated';
                statusDiv.textContent = `Displaying ${images.length} recent recognition results`;
            } else {
                statusDiv.className = 'status waiting';
                statusDiv.textContent = 'Waiting for recognition results...';
            }
            
            // Update each slot
            slots.forEach((slotId, index) => {
                const slot = document.getElementById(slotId);
                
                if (index < images.length) {
                    const image = images[index];
                    slot.className = 'image-slot has-image';
                    slot.innerHTML = `
                        <img src="/annotated/${image.filename}?t=${new Date().getTime()}" 
                             alt="Recognition Result ${index + 1}"
                             onerror="this.style.display='none'">
                        <div class="image-label">${image.label || 'Unknown'}</div>
                        <div class="image-timestamp">${new Date(image.timestamp * 1000).toLocaleString()}</div>
                    `;
                } else {
                    slot.className = 'image-slot';
                    slot.innerHTML = `<div class="empty-slot">Waiting for ${getOrdinal(index + 1)} image...</div>`;
                }
            });
        }
        
        // Helper function to get ordinal numbers
        function getOrdinal(num) {
            const ordinals = ['first', 'second', 'third', 'fourth'];
            return ordinals[num - 1] || `${num}th`;
        }

        // Update dataset statistics
        async function updateDatasetStats() {
            try {
                const response = await fetch('/dataset-stats');
                const data = await response.json();
                
                document.getElementById('totalImages').textContent = data.total_images || 0;
                document.getElementById('totalClasses').textContent = data.total_classes || 0;
                document.getElementById('annotatedImages').textContent = data.annotated_images || 0;
                
                // Update class mapping
                const mappingDiv = document.getElementById('classMapping');
                if (data.class_mapping) {
                    let mappingText = '';
                    for (const [id, name] of Object.entries(data.class_mapping)) {
                        mappingText += `${id}: ${name}\\n`;
                    }
                    mappingDiv.textContent = mappingText || 'No classes defined yet';
                } else {
                    mappingDiv.textContent = 'No classes defined yet';
                }
            } catch (error) {
                console.error('Failed to update dataset stats:', error);
            }
        }
        
        // Handle file upload
        document.getElementById('uploadForm').addEventListener('submit', async function(e) {
            e.preventDefault();
            
            const formData = new FormData(this);
            const uploadResult = document.getElementById('uploadResult');
            const submitBtn = this.querySelector('button[type="submit"]');
            
            submitBtn.disabled = true;
            submitBtn.textContent = 'Processing...';
            uploadResult.innerHTML = '<div style="color: #007bff;">Processing image...</div>';
            
            try {
                const response = await fetch('/image', {
                    method: 'POST',
                    body: formData
                });
                
                const result = await response.json();
                
                if (response.ok) {
                    // Handle both legacy format (predicted_id) and new format (result.image_id)
                    const predictedId = result.predicted_id || result.image_id || result.result?.image_id;
                    const obstacleId = result.obstacle_id || result.result?.obstacle_id;
                    const numObstacles = result.num_obstacles;
                    
                    let successMessage = `
                        <div style="color: #28a745;">
                            <strong>Success!</strong><br>
                            Recognition completed!
                    `;
                    
                    if (numObstacles && numObstacles !== '0') {
                        if (predictedId === '-999') {
                            successMessage += `<br><span style="color: #ffc107;">⚠️ Obstacle ${numObstacles} not saved - No valid detection</span>`;
                        } else {
                            successMessage += `<br><span style="color: #28a745;">✓ Saved as Obstacle ${numObstacles}</span>`;
                            successMessage += `<br><a href="/obstacles" target="_blank" style="color: #007bff; text-decoration: underline;">View Obstacles Display</a>`;
                        }
                    }
                    
                    successMessage += `</div>`;
                    uploadResult.innerHTML = successMessage;
                    
                    // Refresh dataset stats and images immediately after upload
                    updateDatasetStats();
                    setTimeout(checkLatestImages, 500); // Small delay to ensure file is saved
                } else {
                    uploadResult.innerHTML = `<div style="color: #dc3545;">Error: ${result.error}</div>`;
                }
            } catch (error) {
                uploadResult.innerHTML = `<div style="color: #dc3545;">Error: ${error.message}</div>`;
            } finally {
                submitBtn.disabled = false;
                submitBtn.textContent = 'Upload and Predict';
            }
        });
        
        // Download dataset
        async function downloadDataset() {
            const downloadBtn = document.getElementById('downloadBtn');
            const downloadInfo = document.getElementById('downloadInfo');
            
            try {
                downloadBtn.disabled = true;
                downloadBtn.textContent = 'Preparing...';
                downloadInfo.textContent = 'Creating ZIP file, please wait...';
                downloadInfo.style.color = '#007bff';
                
                const response = await fetch('/download-dataset');
                
                if (response.ok) {
                    const blob = await response.blob();
                    const url = window.URL.createObjectURL(blob);
                    const a = document.createElement('a');
                    a.href = url;
                    a.download = `yolo_dataset_${new Date().toISOString().slice(0,10)}.zip`;
                    document.body.appendChild(a);
                    a.click();
                    window.URL.revokeObjectURL(url);
                    document.body.removeChild(a);
                    
                    downloadInfo.textContent = 'Dataset downloaded successfully!';
                    downloadInfo.style.color = '#28a745';
                } else {
                    const errorData = await response.json();
                    downloadInfo.textContent = 'Error: ' + (errorData.message || 'Failed to download');
                    downloadInfo.style.color = '#dc3545';
                }
            } catch (error) {
                downloadInfo.textContent = 'Error: ' + error.message;
                downloadInfo.style.color = '#dc3545';
            } finally {
                downloadBtn.disabled = false;
                downloadBtn.textContent = 'Download Dataset';
                setTimeout(() => { downloadInfo.textContent = ''; }, 5000);
            }
        }
        
        // Initialize
        checkLatestImages();
        updateDatasetStats();
        
        // Auto-refresh every 5 seconds
        setInterval(checkLatestImages, 5000);
        setInterval(updateDatasetStats, 15000);
    </script>
</body>
</html>
"""

# Obstacles display HTML template
OBSTACLES_HTML_TEMPLATE = """
<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>Obstacles Display - YOLO Image Recognition</title>
    <style>
        body {
            font-family: Arial, sans-serif;
            max-width: 1400px;
            margin: 0 auto;
            padding: 20px;
            background-color: #f5f5f5;
        }
        .header {
            text-align: center;
            margin-bottom: 30px;
        }
        .nav {
            text-align: center;
            margin-bottom: 20px;
        }
        .nav a {
            display: inline-block;
            padding: 10px 20px;
            margin: 0 10px;
            background-color: #007bff;
            color: white;
            text-decoration: none;
            border-radius: 5px;
            transition: background-color 0.3s;
        }
        .nav a:hover {
            background-color: #0056b3;
        }
        .obstacles-grid {
            display: grid;
            grid-template-columns: repeat(4, 1fr);
            gap: 20px;
            margin-bottom: 30px;
        }
        .obstacle-card {
            background-color: white;
            border-radius: 10px;
            box-shadow: 0 2px 10px rgba(0,0,0,0.1);
            overflow: hidden;
            position: relative;
        }
        .obstacle-header {
            background-color: #007bff;
            color: white;
            padding: 10px;
            text-align: center;
            font-weight: bold;
        }
        .obstacle-content {
            padding: 15px;
            text-align: center;
            min-height: 280px;
            display: flex;
            flex-direction: column;
            justify-content: center;
            align-items: center;
        }
        .obstacle-image {
            max-width: 100%;
            max-height: 150px;
            border-radius: 5px;
            box-shadow: 0 2px 5px rgba(0,0,0,0.1);
            margin-bottom: 10px;
        }
        .obstacle-placeholder {
            color: #666;
            font-style: italic;
            padding: 40px 20px;
        }
        .obstacle-info {
            margin-top: 10px;
            font-size: 12px;
            color: #666;
        }
        .label-display {
            width: 100%;
            margin-top: 10px;
            padding: 8px;
            background-color: #f8f9fa;
            border: 1px solid #dee2e6;
            border-radius: 4px;
            font-size: 14px;
            min-height: 60px;
        }
        .label-row {
            display: flex;
            justify-content: space-between;
            align-items: center;
            margin-bottom: 5px;
            padding: 2px 0;
        }
        .label-key {
            font-weight: bold;
            color: #495057;
            margin-right: 8px;
        }
        .label-value {
            color: #007bff;
            font-family: monospace;
            background-color: #e9ecef;
            padding: 2px 6px;
            border-radius: 3px;
            font-size: 13px;
        }
        .no-detection {
            color: #6c757d;
            font-style: italic;
        }
        .refresh-btn {
            position: fixed;
            top: 20px;
            right: 20px;
            background-color: #28a745;
            color: white;
            border: none;
            padding: 10px 20px;
            border-radius: 5px;
            cursor: pointer;
            font-size: 14px;
        }
        .refresh-btn:hover {
            background-color: #1e7e34;
        }
        .status {
            text-align: center;
            margin-bottom: 20px;
            padding: 10px;
            background-color: #d4edda;
            color: #155724;
            border-radius: 5px;
        }
    </style>
</head>
<body>
    <button class="refresh-btn" onclick="refreshObstacles()">Refresh</button>
    
    <div class="header">
        <h1>Obstacles Display</h1>
        <p>Real-time display of 8 obstacle recognition results</p>
    </div>
    
    <div class="nav">
        <a href="/">Main Interface</a>
        <a href="/obstacles">Obstacles Display</a>
    </div>
    
    <div class="status" id="status">
        Loading obstacles data...
    </div>
    
    <div class="obstacles-grid" id="obstaclesGrid">
        <!-- Obstacles will be populated here -->
    </div>
    
    <script>
        let obstaclesData = {};
        
        async function loadObstacles() {
            try {
                const response = await fetch('/obstacles-data');
                const data = await response.json();
                obstaclesData = data;
                renderObstacles();
                updateStatus();
            } catch (error) {
                console.error('Error loading obstacles:', error);
                document.getElementById('status').textContent = 'Error loading obstacles data';
                document.getElementById('status').style.backgroundColor = '#f8d7da';
                document.getElementById('status').style.color = '#721c24';
            }
        }
        
        function renderObstacles() {
            const grid = document.getElementById('obstaclesGrid');
            grid.innerHTML = '';
            
            for (let i = 1; i <= 8; i++) {
                const obstacle = obstaclesData[i.toString()];
                const card = document.createElement('div');
                card.className = 'obstacle-card';
                
                const header = document.createElement('div');
                header.className = 'obstacle-header';
                header.textContent = `Obstacle ${i}`;
                
                const content = document.createElement('div');
                content.className = 'obstacle-content';
                
                if (obstacle && obstacle.exists) {
                    const img = document.createElement('img');
                    img.className = 'obstacle-image';
                    img.src = `/obstacles-images/${obstacle.filename}?t=${new Date().getTime()}`;
                    img.alt = `Obstacle ${i}`;
                    img.onerror = function() {
                        this.style.display = 'none';
                        content.innerHTML = '<div class="obstacle-placeholder">Image failed to load</div>';
                    };
                    
                    // Create label display box
                    const labelBox = document.createElement('div');
                    labelBox.className = 'label-display';
                    
                    if (obstacle.label && obstacle.image_id) {
                        labelBox.innerHTML = `
                            <div class="label-row">
                                <span class="label-key">Label:</span>
                                <span class="label-value">${obstacle.label}</span>
                            </div>
                            <div class="label-row">
                                <span class="label-key">ID:</span>
                                <span class="label-value">${obstacle.image_id}</span>
                            </div>
                            <div class="label-row">
                                <span class="label-key">Confidence:</span>
                                <span class="label-value">${(obstacle.confidence * 100).toFixed(1)}%</span>
                            </div>
                        `;
                    } else {
                        labelBox.innerHTML = '<div class="no-detection">No detection data available</div>';
                    }
                    
                    const info = document.createElement('div');
                    info.className = 'obstacle-info';
                    info.textContent = `Last updated: ${new Date(obstacle.timestamp * 1000).toLocaleString()}`;
                    
                    content.appendChild(img);
                    content.appendChild(labelBox);
                    content.appendChild(info);
                } else {
                    const placeholder = document.createElement('div');
                    placeholder.className = 'obstacle-placeholder';
                    placeholder.textContent = 'No image available';
                    
                    const labelBox = document.createElement('div');
                    labelBox.className = 'label-display';
                    labelBox.innerHTML = '<div class="no-detection">Waiting for detection...</div>';
                    
                    content.appendChild(placeholder);
                    content.appendChild(labelBox);
                }
                
                card.appendChild(header);
                card.appendChild(content);
                grid.appendChild(card);
            }
        }
        
        function updateStatus() {
            const existingCount = Object.values(obstaclesData).filter(o => o && o.exists).length;
            const statusEl = document.getElementById('status');
            
            if (existingCount === 0) {
                statusEl.style.backgroundColor = '#fff3cd';
                statusEl.style.color = '#856404';
                statusEl.textContent = 'No obstacles detected yet. Upload images with NUM_OBSTACLES parameter (1-8) and valid detections (ID ≠ -999).';
            } else {
                statusEl.style.backgroundColor = '#d4edda';
                statusEl.style.color = '#155724';
                statusEl.textContent = `Displaying ${existingCount}/8 obstacles with valid detections`;
            }
        }
        
        function refreshObstacles() {
            loadObstacles();
        }
        
        // Initialize
        loadObstacles();
        
        // Auto-refresh every 5 seconds
        setInterval(loadObstacles, 5000);
    </script>
</body>
</html>
"""

# Routes
@app.route('/')
def index():
    """Home page with web interface"""
    return render_template_string(HTML_TEMPLATE)

@app.route('/status', methods=['GET'])
def server_status():
    """Health check endpoint"""
    return jsonify({'status': 'OK'})

@app.route('/upload', methods=['POST'])
def upload_file():
    """Handle file upload and prediction (legacy endpoint - redirects to /image)"""
    # Redirect to /image endpoint to avoid code duplication
    return image_predict()

@app.route('/image', methods=['POST'])
def image_predict():
    """
    This is the main endpoint for the image prediction algorithm
    :return: a json object with a key "result" and value a dictionary with keys "obstacle_id" and "image_id"
    """
    if 'file' not in request.files:
        return jsonify({'error': 'No file part'}), 400
    
    file = request.files['file']
    filename = file.filename
    
    if filename == '':
        return jsonify({'error': 'No selected file'}), 400
    
    # Save to uploads folder first
    file_path = os.path.join(app.config['UPLOAD_FOLDER'], filename)
    file.save(file_path)
    
    # Get parameters from both old and new format
    direction = request.form.get('direction', 'north')
    task_type = request.form.get('task_type', 'TASK_2')
    num_obstacles = request.form.get('NUM_OBSTACLES', '0')  # Support NUM_OBSTACLES parameter
    
    # Try to parse filename format: "<timestamp>_<obstacle_id>_<signal>.jpeg"
    # But be flexible with different formats
    constituents = file.filename.split("_")
    
    # Default values
    obstacle_id = "unknown"
    signal = direction  # Use direction parameter as signal
    
    # Try to extract obstacle_id and signal if available
    try:
        if len(constituents) >= 2:
            obstacle_id = constituents[1]
        if len(constituents) >= 3:
            # Remove file extension from signal
            signal_part = constituents[2]
            # Handle both .jpg and .png extensions
            for ext in ['.jpg', '.jpeg', '.png', '.JPG', '.JPEG', '.PNG']:
                if signal_part.endswith(ext):
                    signal = signal_part[:-len(ext)]
                    break
            else:
                signal = signal_part
    except IndexError:
        # Use default values if parsing fails
        pass
    
    # Check for optional preference parameter
    prefer_close = request.form.get('prefer_close_objects', 'true').lower() == 'true'
    
    # Process the file and predict
    class_id, detection_result = process_file(file_path, signal, task_type, filename)
    
    # Handle NUM_OBSTACLES parameter for obstacle display
    # Only save if detection is valid (image_id != '-999')
    if (num_obstacles and num_obstacles.isdigit() and 1 <= int(num_obstacles) <= 8 
        and detection_result and detection_result.get('image_id') != '-999'):
        save_obstacle_image(detection_result, int(num_obstacles))
        print(f"Obstacle {num_obstacles} saved with valid detection (ID: {detection_result.get('image_id')})")
    elif (num_obstacles and num_obstacles.isdigit() and 1 <= int(num_obstacles) <= 8 
          and detection_result and detection_result.get('image_id') == '-999'):
        print(f"Obstacle {num_obstacles} NOT saved - invalid detection (ID: -999)")
    elif num_obstacles and num_obstacles.isdigit() and 1 <= int(num_obstacles) <= 8:
        print(f"Obstacle {num_obstacles} NOT saved - no detection result")
    
    if detection_result is None:
        return jsonify({'error': 'Failed to process image'}), 500
    
    # Extract image_id from detection result
    image_id = detection_result["image_id"]
    
    print(f"Original image saved to: {detection_result.get('original_image_path', 'N/A')}")
    print(f"Annotated image saved to: {detection_result['marked_image_path']}")
    print(f"YOLO txt file saved to: {detection_result.get('txt_file_path', 'N/A')}")
    
    # Determine response format based on request context
    # If it's from the web interface (has direction/task_type), return simple format
    if direction != 'north' or task_type != 'TASK_1' or num_obstacles != '0':
        # Web interface format
        return jsonify({
            'message': 'File successfully uploaded and processed',
            'predicted_id': image_id,
            'direction': direction,
            'task_type': task_type,
            'num_obstacles': num_obstacles
        }), 200
    else:
        # Original API format
        result = {
            "obstacle_id": obstacle_id,
            "image_id": image_id,
            "detection": {
                "label": detection_result["label"],
                "confidence": detection_result["confidence"],
                "bbox_coordinates": detection_result["bbox"],
                "original_image_path": detection_result.get("original_image_path"),
                "annotated_image_path": detection_result["marked_image_path"],
                "txt_file_path": detection_result.get("txt_file_path")
            }
        }
        return jsonify(result)

@app.route('/latest-result')
def get_latest_result():
    """Get the latest annotated image (legacy endpoint)"""
    if not os.path.exists(ANNOTATED_FOLDER):
        return jsonify({"success": False, "message": "Annotated folder not found"})
    
    # Get all annotated images
    annotated_files = []
    for filename in os.listdir(ANNOTATED_FOLDER):
        if filename.startswith('annotated_') and filename.lower().endswith(('.png', '.jpg', '.jpeg')):
            filepath = os.path.join(ANNOTATED_FOLDER, filename)
            mtime = os.path.getmtime(filepath)
            annotated_files.append((filename, mtime))
    
    if not annotated_files:
        return jsonify({"success": False, "message": "No annotated images found"})
    
    # Sort by modification time, get latest
    annotated_files.sort(key=lambda x: x[1], reverse=True)
    latest_file = annotated_files[0][0]
    
    return jsonify({
        "success": True,
        "image_path": latest_file,
        "timestamp": annotated_files[0][1]
    })

@app.route('/latest-images')
def get_latest_images():
    """Get the latest 4 annotated images with labels"""
    if not os.path.exists(ANNOTATED_FOLDER):
        return jsonify({"success": False, "message": "Annotated folder not found", "images": []})
    
    # Get all annotated images with their info
    annotated_files = []
    for filename in os.listdir(ANNOTATED_FOLDER):
        if filename.startswith('annotated_') and filename.lower().endswith(('.png', '.jpg', '.jpeg')):
            filepath = os.path.join(ANNOTATED_FOLDER, filename)
            mtime = os.path.getmtime(filepath)
            
            # Try to extract label from filename
            # Filename format: annotated_<class_name>_<timestamp>.jpg
            label = 'Unknown'
            try:
                # Remove 'annotated_' prefix and file extension
                base_name = filename[10:]  # Remove 'annotated_'
                if '.' in base_name:
                    base_name = base_name[:base_name.rfind('.')]
                
                # Split by underscore and take first part as class name
                parts = base_name.split('_')
                if len(parts) >= 2:
                    label = parts[0]  # Class name should be first part
                elif len(parts) == 1:
                    label = parts[0]
            except:
                label = 'Unknown'
            
            annotated_files.append({
                'filename': filename,
                'timestamp': mtime,
                'label': label
            })
    
    # Sort by modification time, get latest 4
    annotated_files.sort(key=lambda x: x['timestamp'], reverse=True)
    latest_images = annotated_files[:4]
    
    return jsonify({
        "success": True,
        "images": latest_images
    })

@app.route('/annotated/<filename>')
def serve_annotated_image(filename):
    """Serve annotated images"""
    return send_from_directory(ANNOTATED_FOLDER, filename)

@app.route('/dataset-stats')
def get_dataset_stats():
    """Get dataset statistics"""
    stats = {
        'total_images': 0,
        'total_classes': 0,
        'annotated_images': 0,
        'class_mapping': {}
    }
    
    # Count images
    if os.path.exists(IMAGES_FOLDER):
        stats['total_images'] = len([f for f in os.listdir(IMAGES_FOLDER) if f.lower().endswith(('.png', '.jpg', '.jpeg'))])
    
    # Count annotated images
    if os.path.exists(ANNOTATED_FOLDER):
        stats['annotated_images'] = len([f for f in os.listdir(ANNOTATED_FOLDER) if f.lower().endswith(('.png', '.jpg', '.jpeg'))])
    
    # Load class mapping
    stats['class_mapping'] = load_class_mapping()
    stats['total_classes'] = len(stats['class_mapping'])
    
    return jsonify(stats)

@app.route('/download-dataset')
def download_dataset():
    """Download the complete YOLO dataset as ZIP"""
    if not os.path.exists(DATASET_FOLDER):
        return jsonify({"success": False, "message": "Dataset folder not found"}), 404
    
    # Check if there are files to download
    has_files = False
    for folder in [IMAGES_FOLDER, LABELS_FOLDER, ANNOTATED_FOLDER]:
        if os.path.exists(folder) and os.listdir(folder):
            has_files = True
            break
    
    if not has_files:
        return jsonify({"success": False, "message": "No files found in dataset"}), 404
    
    # Create timestamp for filename
    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
    zip_filename = f"yolo_dataset_{timestamp}.zip"
    zip_path = os.path.join(UPLOAD_FOLDER, zip_filename)
    
    try:
        with zipfile.ZipFile(zip_path, 'w', zipfile.ZIP_DEFLATED) as zipf:
            # Add all files from dataset structure
            for root, dirs, files in os.walk(DATASET_FOLDER):
                for file in files:
                    file_path = os.path.join(root, file)
                    arcname = os.path.relpath(file_path, DATASET_FOLDER)
                    zipf.write(file_path, arcname)
        
        return send_file(zip_path, as_attachment=True, download_name=zip_filename)
    
    except Exception as e:
        return jsonify({"success": False, "message": f"Error creating zip: {str(e)}"}), 500
    
    finally:
        # Clean up temporary file
        try:
            if os.path.exists(zip_path):
                os.remove(zip_path)
        except:
            pass

@app.route('/obstacles')
def obstacles_display():
    """Obstacles display page"""
    return render_template_string(OBSTACLES_HTML_TEMPLATE)

@app.route('/obstacles-data')
def get_obstacles_data():
    """Get obstacles images data with labels and IDs"""
    obstacles_folder = os.path.join(PC_CONFIG.BASE_DIR, "obstacles_display")
    obstacles = {}
    
    if os.path.exists(obstacles_folder):
        for i in range(1, 9):  # 1 to 8
            filename = f"obstacle_{i}.jpg"
            filepath = os.path.join(obstacles_folder, filename)
            if os.path.exists(filepath):
                mtime = os.path.getmtime(filepath)
                
                # Try to find corresponding detection info
                detection_info = get_obstacle_detection_info(i)
                
                obstacles[str(i)] = {
                    'filename': filename,
                    'timestamp': mtime,
                    'exists': True,
                    'label': detection_info.get('label', 'Unknown'),
                    'image_id': detection_info.get('image_id', 'N/A'),
                    'confidence': detection_info.get('confidence', 0.0)
                }
            else:
                obstacles[str(i)] = {
                    'filename': None,
                    'timestamp': None,
                    'exists': False,
                    'label': None,
                    'image_id': None,
                    'confidence': None
                }
    
    return jsonify(obstacles)

@app.route('/obstacles-images/<filename>')
def serve_obstacle_image(filename):
    """Serve obstacle images"""
    obstacles_folder = os.path.join(PC_CONFIG.BASE_DIR, "obstacles_display")
    return send_from_directory(obstacles_folder, filename)

@app.route('/display_stitched', methods=['POST'])
def display_stitched():
    """Display stitched images"""
    try:
        showAnnotatedStitched()
        return jsonify({'display_stitched': 'OK'})
    except Exception as e:
        return jsonify({'error': str(e)}), 500

if __name__ == '__main__':
    print()
    print(f"UPLOAD FOLDER: {UPLOAD_FOLDER}")
    print(f"DATASET FOLDER: {DATASET_FOLDER}")
    print(f"Starting Enhanced Image Recognition Server...")
    print(f"Web interface available at: http://{HOST}:{PORT}")
    print(f"Obstacles display available at: http://{HOST}:{PORT}/obstacles")
    print(f"API endpoints: /image (unified endpoint), /upload (legacy redirect)")
    
    try:
        app.run(host=HOST, port=PORT, debug=False)
    except:
        print('Unable to connect to configured host and port. Switching to localhost:7860.')
        print(f"Web interface: http://localhost:7860")
        print(f"Obstacles display: http://localhost:7860/obstacles")
        app.run(host='0.0.0.0', port=7860, debug=True)