Vision and LiDAR Data Fusion System for Tree Species Detection in an Urban Scale
Abstract
Over the last few decades, urban density has been increasing due to industrial revolution and technology. Therefore, urban areas are suffering from environmental issues at local and global scale. Urban trees are one of the vital aspects of handling these issues. For this, it is required to sense the different attributes of the trees at urban scale. Light Detection and Ranging (LiDAR) gives sparse output with depth values while the RGB image gives high resolution output with no depth values. For better understanding of deep tree features, it is required to utilize both the sensors. Hence, advantages of cameras and LiDAR have been exploited in order to collect the data for accurate and reliable tree species detection. In this study, detection of the tree species has been performed at urban scale using a deep learning-based approach by fusing the LiDAR and camera data. For this study, massive scanning of green space has been done at city scale using RGB-LiDAR system. For data collection of different urban tree species, RGB-LiDAR system has been installed over the car rooftop and the system has been configured with the installed monitor screen inside the car. For training and testing, it is targeted around 400-500 kms to collect the data of different urban tree species around the city. Initially, deep learning-based LiDAR and camera model has been trained individually using urban scale data and then, trained model performs tree species detection at an individual level. Further, decision level fusion strategy has been incorporated for detecting the accurate tree species from an individual levels detected tree species. This work can be extended to develop and maintain tree level database at urban scale.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2020
- Bibcode:
- 2020AGUFMIN007..10L
- Keywords:
-
- 1912 Data management;
- preservation;
- rescue;
- INFORMATICS;
- 1916 Data and information discovery;
- INFORMATICS;
- 1942 Machine learning;
- INFORMATICS;
- 1960 Portals and user interfaces;
- INFORMATICS