Feature-Based Object Detection and Tracking: A Systematic Literature Review
Artikel i vetenskaplig tidskrift, 2023

Correct object detection plays a key role in generating an accurate object tracking result. Feature-based methods have the capability of handling the critical process of extracting features of an object. This paper aims to investigate object tracking using feature-based methods in terms of (1) identifying and analyzing the existing methods; (2) reporting and scrutinizing the evaluation performance matrices and their implementation usage in measuring the effectiveness of object tracking and detection; (3) revealing and investigating the challenges that affect the accuracy performance of identified tracking methods; (4) measuring the effectiveness of identified methods in terms of revealing to what extent the challenges can impact the accuracy and precision performance based on the evaluation performance matrices reported; and (5) presenting the potential future directions for improvement. The review process of this research was conducted based on standard systematic literature review (SLR) guidelines by Kitchenam's and Charters'. Initially, 157 prospective studies were identified. Through a rigorous study selection strategy, 32 relevant studies were selected to address the listed research questions. Thirty-two methods were identified and analyzed in terms of their aims, introduced improvements, and results achieved, along with presenting a new outlook on the classification of identified methods based on the feature-based method used in detection and tracking process.

object tracking

multiple object tracking precision

object detection

object tracking challenges

multiple object tracking accuracy

Feature-based

Författare

Nurul Izzatie Husna Fauzi

Universiti Malaysia Pahang

Zalili Musa

Universiti Malaysia Pahang

Fadhl Mohammad Omar Hujainah

Chalmers, Data- och informationsteknik, Interaktionsdesign och Software Engineering

International Journal of Image and Graphics

0219-4678 (ISSN)

2450037

Ämneskategorier

Annan data- och informationsvetenskap

Programvaruteknik

Datorseende och robotik (autonoma system)

DOI

10.1142/S0219467824500372

Mer information

Senast uppdaterat

2023-03-14