Vehicle detection using OpenCV Python

Vehicle detection is an important application of computer vision that helps in monitoring traffic, automating parking systems and surveillance systems. In this article, we’ll implement a simple vehicle detection system using Python and OpenCV using a pre-trained Haar Cascade classifier and we will get a video in which vehicles will be detected and it will be represented by a rectangular frame around it.

Step 1: Installing OpenCV Library

We need to install the OpenCV library or its implementation. To install it in our system we can use the below pip command:

pip install opencv-python

Step 2: Importing Required Libraries

We will using OpenCV for vehicle detection and handling video streams. Also we need time module to add delays after detecting a vehicle.

import cv2
import time

Step 3: Loading Haar Cascade Classifier and Video

We need to downlaod the Haar Cascade files (download it from here) and load the Haar Cascade classifier that’s trained to detect vehicles. Also we have to load the video file (download it from here) from which vehicles will be detected. Now to capture a video we need to create a VideoCapture() object.

haar_cascade = 'cars.xml'
car_cascade = cv2.CascadeClassifier(haar_cascade)
if car_cascade.empty():
    raise Exception("Error loading Haar cascade file.")

video = 'car_video.avi'
cap = cv2.VideoCapture(video)

if not cap.isOpened():
    raise Exception("Error opening video file.")

Step 4: Detecting Vehicles in Video Frames

We read each frame of the video, convert it to grayscale and use the Haar Cascade classifier to detect vehicles. Once a vehicle is detected we draw rectangles around it.

• cars = car_cascade.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=3): Detects vehicles in the grayscale image using the Haar Cascade classifier.

car_detected = False  
detection_time = None  

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

    gray = cv2.cvtColor(frames, cv2.COLOR_BGR2GRAY)
    cars = car_cascade.detectMultiScale(gray, scaleFactor=1.1, minNeighbors=3)

    
    if len(cars) > 0 and not car_detected:
        car_detected = True
        detection_time = time.time()
        print("Car detected!")

    if car_detected and time.time() - detection_time > 5:  
        break

    for (x, y, w, h) in cars:
        cv2.rectangle(frames, (x, y), (x + w, y + h), (0, 0, 255), 2)

    cv2.imshow('video', frames)

    if cv2.waitKey(33) == 27: 
        break

Step 5: Release Video and Close Windows

After processing the frames we release the video capture object and close any OpenCV windows.

cap.release()  
cv2.destroyAllWindows()

Output:

Vehicle detection systems can be further enhanced by integrating more advanced techniques such as deep learning-based models which offer better accuracy and robustness in complex environments. By using frameworks like TensorFlow or PyTorch vehicle detection systems can achieve real-time performance and handle various challenges.

Note : Above code will not run on online IDE. It will work on local system only.
You can download source code from here.

For more articles on detections using Haar Cascade - Face Detection Basics