How to Track Object in Video Using OpenCV in Computer Vision

To track an object in video using OpenCV, you first select the object region with cv2.selectROI, then create a tracker like cv2.TrackerCSRT_create(), and update the tracker frame-by-frame with tracker.update(). This process allows you to follow the object's position across video frames efficiently.

Syntax

Object tracking in OpenCV involves these main steps:

• cv2.selectROI: Manually select the object region in the first frame.
• cv2.TrackerCSRT_create(): Create a tracker instance (CSRT is accurate and robust).
• tracker.init(frame, bbox): Initialize the tracker with the first frame and bounding box.
• tracker.update(frame): Update the tracker with each new frame to get the new bounding box.

python

import cv2

# Create tracker
tracker = cv2.TrackerCSRT_create()

# Select ROI on first frame
bbox = cv2.selectROI('Frame', frame, False)

# Initialize tracker
tracker.init(frame, bbox)

# In loop for each frame:
(success, bbox) = tracker.update(frame)

Example

This example shows how to track a selected object in a video file or webcam stream using OpenCV's CSRT tracker.

python

import cv2

# Open video or webcam
video = cv2.VideoCapture(0)  # Use 0 for webcam or replace with video file path

# Read first frame
ret, frame = video.read()
if not ret:
    print('Failed to read video')
    exit()

# Select ROI
bbox = cv2.selectROI('Tracking', frame, False)
cv2.destroyWindow('Tracking')

# Create tracker
tracker = cv2.TrackerCSRT_create()
tracker.init(frame, bbox)

while True:
    ret, frame = video.read()
    if not ret:
        break

    # Update tracker
    success, bbox = tracker.update(frame)

    if success:
        # Draw bounding box
        x, y, w, h = map(int, bbox)
        cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 2)
        cv2.putText(frame, 'Tracking', (x, y - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0), 2)
    else:
        cv2.putText(frame, 'Lost', (20, 50), cv2.FONT_HERSHEY_SIMPLEX, 0.75, (0, 0, 255), 2)

    cv2.imshow('Tracking', frame)

    if cv2.waitKey(1) & 0xFF == 27:  # ESC key to exit
        break

video.release()
cv2.destroyAllWindows()

Output

A window opens showing the video stream with a green rectangle tracking the selected object; if tracking fails, 'Lost' appears in red text.

Common Pitfalls

Common mistakes when tracking objects with OpenCV include:

• Not selecting the ROI correctly or forgetting to initialize the tracker.
• Using an incompatible tracker for the object or video conditions (CSRT is good for accuracy, KCF is faster but less robust).
• Not handling tracker failure when the object is lost or occluded.
• Forgetting to release video capture and destroy windows, causing resource leaks.

Always check the success flag from tracker.update() to handle lost tracking gracefully.

python

import cv2

# Wrong: Not initializing tracker before update
tracker = cv2.TrackerCSRT_create()

# This will fail because tracker.init() is missing
ret, frame = video.read()
(success, bbox) = tracker.update(frame)  # Wrong usage

# Correct usage:
# tracker.init(frame, bbox) before calling update

Quick Reference

Step	Function	Description
1	cv2.selectROI(window, frame, False)	Manually select object bounding box
2	cv2.TrackerCSRT_create()	Create a CSRT tracker instance
3	tracker.init(frame, bbox)	Initialize tracker with first frame and bbox
4	success, bbox = tracker.update(frame)	Update tracker for each new frame
5	cv2.rectangle(frame, (x,y), (x+w,y+h), color, thickness)	Draw bounding box on frame

Key Takeaways

Use cv2.selectROI to manually select the object region in the first frame.

Create and initialize a tracker like cv2.TrackerCSRT_create() before updating it.

Call tracker.update(frame) on each new frame to get the object's new position.

Check the success flag from tracker.update() to handle tracking failures.

Release video capture and destroy OpenCV windows to free resources.