Abstract. During the last decade, the use of machine and deep learning tools to support 3D semantic segmentation of point clouds remarkably increased and their impressive results have led to the application of such methods to the semantic modeling of heritage buildings. Nevertheless, a standard procedure to deal with such problem is still missing, and several significant challenges, caused by the complexity of heritage building scenario, have still to be faced. This paper aims at comparing the overall performance of two convolutional neural network architectures, named SegNet and Deeplabv3+, for the semantic segmentation of heritage point clouds throughout a multiview approach. More specifically, the two architectures have been tested to obtain 2D segmentation maps of the related photogrammetric images of the buildings, and then the output maps have been projected to the photogrammetric point cloud by means of the interior and exterior camera parameters. Experiments to test the effectiveness of the proposed approach have been conducted on the case study of Spedale del Ceppo in Pistoia, Italy. Despite the results shown a remarkable performance of both the architectures, Deeplabv3+ outperformed SegNet in terms of accuracy, memory consumption and training time.
Abstract. The development of remote sensing techniques dramatically improved the human knowledge of natural phenomena and the real time monitoring and interpretation of the events happening in the environment. The recently developed terrestrial, aerial and satellite remote sensors caused the availability of huge amount of data. The large size of such data is leading the research community to the search for efficient methods for real time information extraction, and, more in general, understanding the collected data. Nowadays, this is typically done by means of artificial intelligence-based methods, and, more specifically, usually by means of machine learning tools. Focusing on semantic segmentation, which is clearly related to a proper interpretation of the acquired remote sensing data, supervised machine learning is often used: it is based on the availability of a set of ground truth labeled data, which are used in order to properly train a machine learning classifier. Despite the latter, after a proper training phase, usually allows to obtain quite effective segmentation results, the ground truth labeled data production is usually a very laborious and time consuming task, performed by a human operator. Motivated by the latter consideration, this work aims at introducing a graphical interface developed in order to support semi-automatic semantic segmentation of images acquired by a UAS. Certain of the potentialities of the proposed graphical are shown in the specific case of plastic litter detection in multi-spectral images.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.