Objective Construction of Ground Truth Images

Smith, Mark and Maiti, Ananda and Maxwell, Andrew and Kist, Alexander A. ORCID: https://orcid.org/0000-0001-9105-7050 (2021) Objective Construction of Ground Truth Images. In: 17th International Conference on Remote Engineering and Virtual Instrumentation (REV 2020), 26 Feb - 28 Feb 2020, Athens, United States.


Abstract

In the context of virtual and augmented reality, computer vision plays a pivotal role. To benchmark performance, evaluation of computer vision models, such as edge detection is essential. Traditionally this has relied on subjective analysis of the resultant images. Alternatively, models have been assessed against ground truth images. However, ground truth images are highly subjective, relying on human judging to determine the appropriate location of features. Literature complains about the lack of objective quantitative measures for model evaluation, yet no solution has been presented. Ground truth is the objective verification of properties of an image. In the context of this paper it is a data set that includes an accurate and complete representation of the edges. The subjective nature of creating ground truth images has meant that true image analysis model evaluation has been limited. Reducing the level of subjective decisions can improve the confidence level when measuring the performance of computer vision image analysis models. This work describes a new method to improve ground truth image confidence through an automated computer vision feature detection model voting system.


Statistics for USQ ePrint 47147
Statistics for this ePrint Item
Item Type: Conference or Workshop Item (Commonwealth Reporting Category E) (Paper)
Refereed: Yes
Item Status: Live Archive
Faculty/School / Institute/Centre: Historic - Faculty of Health, Engineering and Sciences - School of Mechanical and Electrical Engineering (1 Jul 2013 - 31 Dec 2021)
Faculty/School / Institute/Centre: Historic - Faculty of Health, Engineering and Sciences - School of Mechanical and Electrical Engineering (1 Jul 2013 - 31 Dec 2021)
Date Deposited: 05 Apr 2022 02:39
Last Modified: 31 May 2022 00:27
Uncontrolled Keywords: Computer vision; Edge detection; Ground truth
Fields of Research (2020): 46 INFORMATION AND COMPUTING SCIENCES > 4603 Computer vision and multimedia computation > 460306 Image processing
46 INFORMATION AND COMPUTING SCIENCES > 4607 Graphics, augmented reality and games > 460799 Graphics, augmented reality and games not elsewhere classified
Socio-Economic Objectives (2020): 28 EXPANDING KNOWLEDGE > 2801 Expanding knowledge > 280110 Expanding knowledge in engineering
Identification Number or DOI: https://doi.org/10.1007/978-3-030-52575-0_24
URI: http://eprints.usq.edu.au/id/eprint/47147

Actions (login required)

View Item Archive Repository Staff Only