Please use this identifier to cite or link to this item:
https://scholarbank.nus.edu.sg/handle/10635/224136
Title: | UNCERTAINTY CONTROL FOR ARTIFICIAL INTELLIGENCE (AI) � BASED FAÇADE VISUAL INSPECTION | Authors: | LEE JIN NING DEANNA | Keywords: | 2020-2021 Building Bachelor's BACHELOR OF SCIENCE (PROJECT AND FACILITIES MANAGEMENT) Wang Qian |
Issue Date: | 7-May-2021 | Citation: | LEE JIN NING DEANNA (2021-05-07). UNCERTAINTY CONTROL FOR ARTIFICIAL INTELLIGENCE (AI) � BASED FAÇADE VISUAL INSPECTION. ScholarBank@NUS Repository. | Abstract: | There is a pressing need to move from traditional visual faced inspection as it is costly, dangerous, lacks efficiency and especially with the new changes made to the building control act in 2020 which requires more visual facade inspection to be conducted. AI-based visual facade inspection will be the solution for the above-mentioned problem. However, development of DL model demands great human involvement in cognition, decision-making and implementation of activities. Poor human behaviors could impact reliability and accuracy of AI-based inspection. As this technology is still relatively new in the market, there is lack of quantitative representation to quantify influence of human factors on AI-based façade visual inspection’s reliability and efficiency. There is also lack of data for apprehending human activities in the highly unpredictable AI-based façade visual inspection and difficulty in accurately translating biased prone human opinions into qualitative results. Hence, the focus of this dissertation is to pinpoint critical activities affecting reliability and efficiency of the DL model of AI-based façade visual inspection using modified FDM, a novel methodology created by us. Through analyst of weighted evaluation results such as defuzzification, COV and stability of the results of the FDM questionnaires, 12 UE (Uncertainty of Efficiency) activities and 18 UR (Uncertainty of Reliability) activities were identified to be critical factors that should be further studied to further improve the DL model to a confidence level feasible for major deployment. | URI: | https://scholarbank.nus.edu.sg/handle/10635/224136 |
Appears in Collections: | Bachelor's Theses |
Show full item record
Files in This Item:
File | Description | Size | Format | Access Settings | Version | |
---|---|---|---|---|---|---|
Lee Jin Ning Deanna 2020-2021_dissertation.pdf.pdf | 3.46 MB | Adobe PDF | RESTRICTED | None | Log In |
Google ScholarTM
Check
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.