Real-Time Object Detection, Tracking, and Facial Analysis

This Python script performs real-time object detection using YOLOv8, tracks detected objects, and analyzes faces for emotion and age estimation. It integrates multiple models for comprehensive scene understanding.

Features

• Object Detection: Uses YOLOv8 to detect objects from the COCO dataset.
• Object Tracking: Implements OpenCV trackers (CSRT/KCF) to follow detected objects.
• Facial Analysis:
  • Face detection using dlib.
  • Emotion recognition (Angry, Disgust, Fear, Happy, Sad, Surprise, Neutral).
  • Age group estimation (e.g., 0-2, 4-6, 15-20, etc.).
• Real-Time Webcam Support: Works with default webcam or video files.

Prerequisites

• Python 3.8+
• Required files (place in the project directory):
  • yolov8n.pt (YOLOv8 model)
  • coco.names (COCO class labels)
  • emotion_model.h5 (Emotion recognition model)
  • age_model.h5 (Age estimation model)
  • shape_predictor_68_face_landmarks.dat (dlib face landmark detector)

1. Install dependencies:

   pip install opencv-python opencv-contrib-python numpy keras ultralytics dlib

   Note: For dlib installation issues on Windows, refer to dlib's official guide.

Usage

1. Place all required files (listed in prerequisites) in the project directory.

2. Run the script:

   python object_detection_tracking.py

3. Press q to exit the application.

Using a Video File

Modify line 102 in the script:

cap = cv2.VideoCapture("path/to/your/video.mp4")  # Replace 0 with video path

Key Functions

• detect_objects(): Detects objects using YOLOv8.
• track_objects(): Tracks detected objects using OpenCV trackers.
• predict_emotion_and_age(): Predicts emotion and age group for detected faces.

Notes

• Tracker Selection: Switch between TrackerCSRT (high accuracy) and TrackerKCF (fast) in create_tracker().
• Debug Outputs: Emotion/age prediction confidence scores are printed in the console.
• Performance: Processing speed depends on hardware. Reduce input resolution if lag occurs.

Troubleshooting

1. Missing Files Error:

   • Ensure all required files are in the project directory.
   • Download links:
     • YOLOv8 Model
     • dlib Shape Predictor
     • COCO names, emotion, and age models (custom training or third-party sources).

2. Model Loading Errors:

   • Verify Keras model compatibility (custom objects are defined in the script).
   • Ensure compile=False is retained when loading models.

3. Webcam Not Working:

   • Change cv2.VideoCapture(0) to use a different camera index (e.g., 1).

Sample Output

• Bounding boxes around detected objects (green).
• Emotion and age labels above detected faces.
• Real-time display in a window titled "Frame".

---