tracking

Question 1

What is the goal of object tracking in video?

Answer 1

To determine the location of a target object in each frame of a video sequence.

Question 3

What is change detection used for in tracking?

Answer 3

To identify moving objects in static camera scenes by detecting changes from a reference background.

Question 5

What is simple frame differencing?

Answer 5

Computing the difference between consecutive frames and thresholding the result to detect motion.

Question 7

What is a limitation of frame differencing?

Answer 7

It’s sensitive to noise and changes in lighting.

Question 9

What is background modeling in tracking?

Answer 9

Building a model of the scene’s static content and identifying pixels that differ as foreground.

Question 11

What is a moving median background model?

Answer 11

A model that stores the median of recent frames at each pixel to filter out short-term variations.

Question 13

What is the drawback of moving median background modeling?

Answer 13

It’s computationally expensive and may still struggle with complex backgrounds.

Question 15

What is the key idea behind Gaussian Mixture Models in tracking?

Answer 15

Each pixel is modeled as a mixture of Gaussians over time, capturing dynamic background behavior.

Question 17

What does each Gaussian component in a GMM represent?

Answer 17

A different background or foreground mode for a pixel’s intensity over time.

Question 19

How are pixels classified in GMM-based background subtraction?

Answer 19

Pixels fitting the dominant Gaussian are background; others are foreground.

Question 21

Who introduced the GMM method for background subtraction?

Answer 21

Stauffer and Grimson (1998).

Question 23

What is template-based tracking?

Answer 23

Matching an object’s appearance or histogram from earlier frames to its new location in the current frame.

Question 25

What are the two main types of template-based tracking?

Answer 25

Appearance-based and histogram-based tracking.

Question 27

What does appearance-based template matching use?

Answer 27

Image patches matched using cross-correlation or squared difference.

Question 29

What is a limitation of appearance-based template matching?

Answer 29

It's sensitive to scale, pose, and occlusion.

Question 31

What does histogram-based template matching use?

Answer 31

Color histograms of the object region, often with weighting.

Question 33

What is a benefit of histogram-based tracking?

Answer 33

It's more robust to lighting changes, pose variations, and occlusion.

Question 35

What is a weighted histogram in tracking?

Answer 35

A histogram where central pixels of the object receive more weight to improve accuracy.

Question 37

What is feature-based tracking?

Answer 37

Detecting and matching keypoints across frames to locate and track an object.

Question 39

What types of features are used for feature-based tracking?

Answer 39

Corners, blobs, and invariant descriptors like SIFT, SURF, or ORB.

Question 41

What are the advantages of feature-based tracking?

Answer 41

It's robust to scale, rotation, and partial occlusion.

Question 43

What is the typical pipeline for feature-based tracking?

Answer 43

1. Detect features, 2. Match features between frames, 3. Track region with most consistent matches.

Question 45

What is an example of a large-scale tracking method?

Answer 45

Benfold (2011) — multi-person tracking using feature trajectories in real scenes.

Question 47

Which tracking method is best for dynamic backgrounds?

Answer 47

GMM-based background subtraction.

Question 49

Which tracking method is fastest but least robust?

Answer 49

Simple frame differencing.

Question 51

Which tracking method is best for complex scenes with occlusion and scale change?

Answer 51

Feature-based tracking.

Question 53

Why is object tracking important in computer vision?

Answer 53

It's essential for tasks like surveillance, video analysis, robotics, and augmented reality.