| Title | A Preliminary Study on the Deep Learning Applied to Geomorphological Identification of Slope by Multi-Phase and Multi- Scale Images (2/2) - Probe on Image Processing and Applicability |
| Dept | Transportation Technology Research Center |
| Year | 2024 |
| Month | 3 |
| Price | 200 |
| Summary | One of the annual policies focuses of Ministry of Transportation and Communication’s administrative plan is to “strengthen intelligent applications and increase transportation efficiencies”, which specifies intensified application of new-emerging technologies like 5G and AI to realize transformation of digitized and intelligent traffic. The 2020 Transportation Policy White Paper states the “application of AI, UAV and remote sensing technology in railway/highway inspection or monitoring” as a key action program. Therefore, the project of "A Preliminary Study on the Deep Learning Applied to Geomorphological Identification of Slope by Multi-Phase and Multi-Scale Images" starting from 2022. The purpose of the project is to explore the combination of multi-period and multi-scale aerial images (such as satellites, aerial photos, UAV) and deep learning technology to explore the application to geomorphological identification of highway slopes. The present research being the second year of a project that spans two years, the main research results: (1) the collection of documents on deep learning applied to image recognition technology and the probe in the methods and feasibility of image recognition from geomorphological identification on slopes; (2) the analyzation of multi-scale remote sensing such as satellites, aerial surveys, UAVs, etc. and aerial photograph processing methods; (3) the interview with end-user to understand the maintenance of slope and the demand; and (4) the explanation of the deep learning neural network model that apply to landform identification. Benefits: The present research probe the feasibility of the application of AI technology in combination with multi-phase and multi-scale images of slope monitoring to perform geomorphological identification, hoping that by the exploration of new-emerging technologies and techniques, utilities such as land feature identification, automatic selection of ranges of bare collapsed land and earth volume and range identification can be achieved in the future to facilitate routine inspection by highway slope administrations and to make recovery works from disasters easier and faster. Applications provided to governments: The results can be provided for Directorate General of Highways and National Freeway Bureau to apply in managing highway slopes and disaster prevention, as well as for the Institute in further probe and application in research and development of AI techniques. |
| Post date | 2024/03/28 |
View count:
131