Object Detection with YOLOv8 and Webcam Feed

This guide explains how to build an object detection system using the YOLOv8 model to detect cigarettes from a webcam feed, with an additional feature of triggering an alert sound when a cigarette is detected.

Requirements

To get started, you need the following libraries and tools:

• Python 3.x
• OpenCV for webcam feed and image processing
• Ultralytics YOLOv8 for object detection
• Pygame to play alert sounds
• Threading for asynchronous sound playback

You can install the required libraries using:

pip install opencv-python ultralytics pygame

Code Overview

Here's the full code for the object detection system:

import cv2
from ultralytics import YOLO
import pygame
import threading
import time

# YOLOv8 model loading
model = YOLO('trained_model.pt')

# Opening the webcam
cap = cv2.VideoCapture(0)

# Alert sound file
alert_sound = 'alert_sound.mp3'

# Initializing pygame for sound playback
pygame.mixer.init()

# Function to play alert sound asynchronously
def play_alert_sound():
    while True:
        pygame.mixer.music.load(alert_sound)
        pygame.mixer.music.play()
        print("Alert sound played!")
        time.sleep(1)

# Starting the sound playback in a new thread
alert_thread = threading.Thread(target=play_alert_sound, daemon=True)
alert_thread.start()

# Main loop for video processing and object detection
while True:
    ret, frame = cap.read()
    if not ret:
        break

    # Detect objects using YOLOv8
    results = model(frame)

    cigarette_detected = False

    for result in results[0].boxes:
        class_id = int(result.cls.item())
        if model.names[class_id] == 'cigarette':
            cigarette_detected = True
            break

    if cigarette_detected:
        print("Cigarette detected!")
        pygame.mixer.music.set_volume(1.0)
    else:
        print("No cigarette detected!")
        pygame.mixer.music.set_volume(0.0)

    annotated_frame = results[0].plot()
    cv2.imshow('Webcam Feed', annotated_frame)

    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

cap.release()
cv2.destroyAllWindows()

How It Works

1. YOLOv8 Model: The YOLOv8 model is pre-trained on various objects, including a cigarette. The trained_model.pt file is the custom-trained YOLOv8 model that detects the "cigarette" class.

2. Webcam Feed: OpenCV is used to capture frames from the webcam. The frames are passed to the YOLOv8 model for object detection.

3. Alert Sound: When a cigarette is detected, the program triggers an alert sound using Pygame. A separate thread is used to play the sound asynchronously, preventing it from blocking the video processing.

4. Real-Time Detection: As the program processes each frame, it checks whether a cigarette is detected. If detected, an alert sound is played, and if no cigarette is found, the sound is muted.

5. Exit Condition: Press the 'q' key to exit the video feed.

Result

The system will show the webcam feed with real-time object detection annotations, and the alert sound will play whenever a cigarette is detected. Here’s an example of what the output might look like:

https://www.youtube.com/watch?v=9ECnFEY9Sg0

Conclusion

This project demonstrates how to integrate object detection with real-time video processing and alert systems. By using YOLOv8 for object detection and Pygame for sound management, we can create an intelligent system that reacts to specific objects detected in the video feed.

---

Tips and Troubleshooting

• Ensure that your webcam is properly connected and accessible by OpenCV.
• If the sound isn't playing, check that your audio settings are correctly configured in Pygame.

---

This is an example of how you can structure and document your work using Markdown. You can include images, like the webcam feed screenshot shown above, and even add links to relevant resources or repositories.

Let me know if you'd like more details or need help with any section!