Building a Real-Time Vehicle Detection and Tracking System on Your Smartphone

Introduction Welcome back! In the first part, we discussed the basics of speed measurement and the technologies involved. Now, we will move on to the practical implementation of detecting and tracking vehicles using YOLOv8 and DeepSORT.

1. Installing and Setting Up YOLOv8 First, you need to install the YOLOv8 model. YOLOv8 can be installed via the Ultralytics library.


from ultralytics import YOLO

# Load a pre-trained YOLOv8 model
model = YOLO('yolov8n.pt')

2. Capturing Video from Your Smartphone You can use OpenCV to capture video from your smartphone’s camera.


import cv2

# Capture video from the smartphone camera
cap = cv2.VideoCapture(0)  # 0 for default camera

while cap.isOpened():
    ret, frame = cap.read()
    if not ret:
        break

    # Process the frame
    results = model.predict(source=frame)
    # Further processing...

3. Implementing Object Tracking with DeepSORT DeepSORT helps in maintaining the identity of vehicles across multiple frames. Here’s how you can integrate it:


from deep_sort_pytorch.deep_sort import DeepSort

# Initialize DeepSORT
deepsort = DeepSort('deep_sort_pytorch/deep/checkpoint/ckpt.t7')

while cap.isOpened():
    ret, frame = cap.read()
    if not ret:
        break

    results = model.predict(source=frame)
    bboxes = extract_bboxes(results)  # Custom function to extract bounding boxes

    # Update tracker
    outputs = deepsort.update(bboxes)
    for output in outputs:
        bbox = output[:4]
        track_id = output[4]
        # Draw bounding box and track ID on frame

4. Displaying the Results Use OpenCV to display the video with detected and tracked vehicles.


    for output in outputs:
        bbox = output[:4]
        track_id = output[4]

        # Draw bounding box and track ID on frame
        cv2.rectangle(frame, (bbox[0], bbox[1]), (bbox[2], bbox[3]), (255, 0, 0), 2)
        cv2.putText(frame, f'ID: {track_id}', (bbox[0], bbox[1]-10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 0, 0), 2)

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

cap.release()
cv2.destroyAllWindows()

Conclusion In this part, we implemented the basics of vehicle detection and tracking using YOLOv8 and DeepSORT. In the final part, we will focus on calculating the speed of the detected vehicles and integrating GPS data for more accurate measurements.