Ubels, Nick (2019) Classifying asbestos roofs in the Dutch province of Drenthe using hyperspectral imagery and deep learning. Bachelor's Thesis, Artificial Intelligence.
|
Text
AI_BA_2019_NickUbels.pdf Download (6MB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
toestemming.pdf Restricted to Registered users only Download (119kB) |
Abstract
Asbestos was used a lot due to its favourable characteristics, but after discovering the health impact of the fibres it was widely banned. Not all asbestos is already gone however, it is still used on various roofs spread across The Netherlands. These roofs need to be remediated before a proposed nationwide ban goes into effect. This study focusses on the province of Drenthe in the north of The Netherlands. Hyperspectral aerial imagery is used to classify buildings as asbestos, clean or suspicious. In an earlier study this has been done using Spectral Angle Mapping, and this is used as a baseline. In this study two deep neural networks, U-Net and DeepLabv3+, are compared to Spectral Angle Mapping to determine if deep learning is a feasible alternative. The networks are trained on three different datasets varying in quality and size. After testing it is found that DeepLabv3+ has better performance on this problem than Spectral Angle Mapping. DeepLabv3+ is able to obtain a mean Intersection over Union of 0.41 versus 0.32 for Spectral Angle Mapping.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Supervisor name: | Wiering, M.A. |
| Degree programme: | Artificial Intelligence |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 31 Jan 2019 |
| Last Modified: | 01 Feb 2019 10:57 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/19092 |
Actions (login required)
 |
View Item |