A Study on Development of Test and Evaluation Methods for AI-based Image Surveillance Systems in Defense
Qupyo Cho, Manho Han, Hongryul Ryu, Jongkwan Lee, Seungmin Kim · Journal of Applied Reliability · 2023
Purpose: This study proposes a method of data set collection and utilization in conjunction with the pathway of test and evaluation (T&E) methods for the military use of AI-based image surveillance systems, which is a crucial application of AI technology. Methods: The inadequacy of current T&E methods used in AI-based image surveillance system in the military was revealed by indicating the incompatibility of current acquisition process and identifying characteristics of the AI-based image surveillance systems for military purposes. Results: A data set collection method as well as alternative T&E methods and an acquisition procedure for the military use of the AI-based image surveillance system are proposed. Conclusion: It is expected that the consensus can be reached for the importance of the development of the T&E methods along with the improvement of the relevant policies and regulations. Moreover, it would serve as a guideline for setting up the procedure of the T&E methods for the AI-based image surveillance system in defense.