[4D3-E-2] Machine learning: living environment

[4D3-E-2-02] Trees Detection on Google Street View Images Using Deep Learning and City Open Data

〇Lieu-Hen Chen¹, Hao-Ming Hung¹, Cheng-Yu Sun¹, Eric Hsiao-Kuang Wu², Toru Yamaguchi³, Eri Sato-Shimokawara³, Hao Chen¹ (1. Nantional Chi Nan University, 2. National Central University, 3. Tokyo Metropolitan University)

キーワード：Google Street View, Open Data, Street Tree, Object Recognition, Object Segmentation

For almost every cities and towns, street trees play an important role in representing seasonal change of the street view. Nowadays, lots of countries start promoting open data. Among these data, very useful information related to street trees are well documented with free access by many city governments. At the same time, Google Street View provides the view of a certain surrounding by composing stitched images which are shot by specialized vehicles moving along streets and alleys. However, few research reports have been published on utilizing city open data for trees detection on Google Street View. Therefore, in this study, we aim to perform trees detection on Google Street View Images by utilizing Deep Learning technologies and city open data.