Classification of Point Clouds for Indoor Components Using Few Labeled Samples
The existing deep learning methods for point cloud classification are trained using abundant labeled samples and used to test only a few samples. However, classification tasks are diverse, and not all tasks have enough labeled samples for training. In this paper, a novel point cloud classification m...
Main Authors: | , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2020-07-01
|
Series: | Remote Sensing |
Subjects: | |
Online Access: | https://www.mdpi.com/2072-4292/12/14/2181 |
id |
doaj-b7c20da07d0644faadd30712bba036f4 |
---|---|
record_format |
Article |
spelling |
doaj-b7c20da07d0644faadd30712bba036f42020-11-25T03:45:20ZengMDPI AGRemote Sensing2072-42922020-07-01122181218110.3390/rs12142181Classification of Point Clouds for Indoor Components Using Few Labeled SamplesHangbin Wu0Huimin Yang1Shengyu Huang2Doudou Zeng3Chun Liu4Hao Zhang5Chi Guo6Long Chen7College of Surveying and Geo-informatics, Tongji University, Shanghai, 200092, ChinaCollege of Surveying and Geo-informatics, Tongji University, Shanghai, 200092, ChinaInstitute of Geodesy and Photogrammetry, Swiss Federal Institute of Technology Zurich, CH-8093 Zurich, SwitzerlandCollege of Surveying and Geo-informatics, Tongji University, Shanghai, 200092, ChinaCollege of Surveying and Geo-informatics, Tongji University, Shanghai, 200092, ChinaSchool of Electronics and Information Engineering, Tongji University, Shanghai 200092, ChinaNational satellite positioning system engineering technology research center, Wuhan University, Wuhan 430072, ChinaSchool of Data and Computer Science, Sun Yat-sen University, Guangzhou 510006, ChinaThe existing deep learning methods for point cloud classification are trained using abundant labeled samples and used to test only a few samples. However, classification tasks are diverse, and not all tasks have enough labeled samples for training. In this paper, a novel point cloud classification method for indoor components using few labeled samples is proposed to solve the problem of the requirement for abundant labeled samples for training with deep learning classification methods. This method is composed of four parts: mixing samples, feature extraction, dimensionality reduction, and semantic classification. First, the few labeled point clouds are mixed with unlabeled point clouds. Next, the mixed high-dimensional features are extracted using a deep learning framework. Subsequently, a nonlinear manifold learning method is used to embed the mixed features into a low-dimensional space. Finally, the few labeled point clouds in each cluster are identified, and semantic labels are provided for unlabeled point clouds in the same cluster by a neighborhood search strategy. The validity and versatility of the proposed method were validated by different experiments and compared with three state-of-the-art deep learning methods. Our method uses fewer than 30 labeled point clouds to achieve an accuracy that is 1.89%–19.67% greater than existing methods. More importantly, the experimental results suggest that this method is not only suitable for single-attribute indoor scenarios but also for comprehensive complex indoor scenarios.https://www.mdpi.com/2072-4292/12/14/2181few labeled samplespoint cloudsclassificationindoor scenario |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Hangbin Wu Huimin Yang Shengyu Huang Doudou Zeng Chun Liu Hao Zhang Chi Guo Long Chen |
spellingShingle |
Hangbin Wu Huimin Yang Shengyu Huang Doudou Zeng Chun Liu Hao Zhang Chi Guo Long Chen Classification of Point Clouds for Indoor Components Using Few Labeled Samples Remote Sensing few labeled samples point clouds classification indoor scenario |
author_facet |
Hangbin Wu Huimin Yang Shengyu Huang Doudou Zeng Chun Liu Hao Zhang Chi Guo Long Chen |
author_sort |
Hangbin Wu |
title |
Classification of Point Clouds for Indoor Components Using Few Labeled Samples |
title_short |
Classification of Point Clouds for Indoor Components Using Few Labeled Samples |
title_full |
Classification of Point Clouds for Indoor Components Using Few Labeled Samples |
title_fullStr |
Classification of Point Clouds for Indoor Components Using Few Labeled Samples |
title_full_unstemmed |
Classification of Point Clouds for Indoor Components Using Few Labeled Samples |
title_sort |
classification of point clouds for indoor components using few labeled samples |
publisher |
MDPI AG |
series |
Remote Sensing |
issn |
2072-4292 |
publishDate |
2020-07-01 |
description |
The existing deep learning methods for point cloud classification are trained using abundant labeled samples and used to test only a few samples. However, classification tasks are diverse, and not all tasks have enough labeled samples for training. In this paper, a novel point cloud classification method for indoor components using few labeled samples is proposed to solve the problem of the requirement for abundant labeled samples for training with deep learning classification methods. This method is composed of four parts: mixing samples, feature extraction, dimensionality reduction, and semantic classification. First, the few labeled point clouds are mixed with unlabeled point clouds. Next, the mixed high-dimensional features are extracted using a deep learning framework. Subsequently, a nonlinear manifold learning method is used to embed the mixed features into a low-dimensional space. Finally, the few labeled point clouds in each cluster are identified, and semantic labels are provided for unlabeled point clouds in the same cluster by a neighborhood search strategy. The validity and versatility of the proposed method were validated by different experiments and compared with three state-of-the-art deep learning methods. Our method uses fewer than 30 labeled point clouds to achieve an accuracy that is 1.89%–19.67% greater than existing methods. More importantly, the experimental results suggest that this method is not only suitable for single-attribute indoor scenarios but also for comprehensive complex indoor scenarios. |
topic |
few labeled samples point clouds classification indoor scenario |
url |
https://www.mdpi.com/2072-4292/12/14/2181 |
work_keys_str_mv |
AT hangbinwu classificationofpointcloudsforindoorcomponentsusingfewlabeledsamples AT huiminyang classificationofpointcloudsforindoorcomponentsusingfewlabeledsamples AT shengyuhuang classificationofpointcloudsforindoorcomponentsusingfewlabeledsamples AT doudouzeng classificationofpointcloudsforindoorcomponentsusingfewlabeledsamples AT chunliu classificationofpointcloudsforindoorcomponentsusingfewlabeledsamples AT haozhang classificationofpointcloudsforindoorcomponentsusingfewlabeledsamples AT chiguo classificationofpointcloudsforindoorcomponentsusingfewlabeledsamples AT longchen classificationofpointcloudsforindoorcomponentsusingfewlabeledsamples |
_version_ |
1724510114656813056 |