Transfer Learning Based on Multi-Branch Architecture Feature Extractor for Airborne LiDAR Point Cloud Semantic Segmentation with Few Samples

The existing deep learning-based Airborne Laser Scanning (ALS) point cloud semantic segmentation methods require a large amount of labeled data for training, which is not always feasible in practice. Insufficient training data may lead to over-fitting. To address this issue, we propose a novel Multi...

Full description

Saved in:
Bibliographic Details
Main Authors: Jialin Yuan, Hongchao Ma, Liang Zhang, Jiwei Deng, Wenjun Luo, Ke Liu, Zhan Cai
Format: Article
Language:English
Published: MDPI AG 2025-07-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/17/15/2618
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849406154319331328
author Jialin Yuan
Hongchao Ma
Liang Zhang
Jiwei Deng
Wenjun Luo
Ke Liu
Zhan Cai
author_facet Jialin Yuan
Hongchao Ma
Liang Zhang
Jiwei Deng
Wenjun Luo
Ke Liu
Zhan Cai
author_sort Jialin Yuan
collection DOAJ
description The existing deep learning-based Airborne Laser Scanning (ALS) point cloud semantic segmentation methods require a large amount of labeled data for training, which is not always feasible in practice. Insufficient training data may lead to over-fitting. To address this issue, we propose a novel Multi-branch Feature Extractor (MFE) and a three-stage transfer learning strategy that conducts pre-training on multi-source ALS data and transfers the model to another dataset with few samples, thereby improving the model’s generalization ability and reducing the need for manual annotation. The proposed MFE is based on a novel multi-branch architecture integrating Neighborhood Embedding Block (NEB) and Point Transformer Block (PTB); it aims to extract heterogeneous features (e.g., geometric features, reflectance features, and internal structural features) by leveraging the parameters contained in ALS point clouds. To address model transfer, a three-stage strategy was developed: (1) A pre-training subtask was employed to pre-train the proposed MFE if the source domain consisted of multi-source ALS data, overcoming parameter differences. (2) A domain adaptation subtask was employed to align cross-domain feature distributions between source and target domains. (3) An incremental learning subtask was proposed for continuous learning of novel categories in the target domain, avoiding catastrophic forgetting. Experiments conducted on the source domain consisted of DALES and Dublin datasets and the target domain consists of ISPRS benchmark dataset. The experimental results show that the proposed method achieved the highest OA of 85.5% and an average F1 score of 74.0% using only 10% training samples, which means the proposed framework can reduce manual annotation by 90% while keeping competitive classification accuracy.
format Article
id doaj-art-8b4f26f3f5984ef89711d656aaaa650f
institution Kabale University
issn 2072-4292
language English
publishDate 2025-07-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj-art-8b4f26f3f5984ef89711d656aaaa650f2025-08-20T03:36:30ZengMDPI AGRemote Sensing2072-42922025-07-011715261810.3390/rs17152618Transfer Learning Based on Multi-Branch Architecture Feature Extractor for Airborne LiDAR Point Cloud Semantic Segmentation with Few SamplesJialin Yuan0Hongchao Ma1Liang Zhang2Jiwei Deng3Wenjun Luo4Ke Liu5Zhan Cai6School of Remote Sensing and Information Engineering, Wuhan University, Wuhan 430072, ChinaSchool of Remote Sensing and Information Engineering, Wuhan University, Wuhan 430072, ChinaFaculty of Resources and Environmental Science, Hubei University, Wuhan 430062, ChinaTianjin Key Laboratory of Rail Transit Navigation Positioning and Spatio-Temporal Big Data Technology, Tianjin 300251, ChinaSchool of Remote Sensing and Information Engineering, Wuhan University, Wuhan 430072, ChinaSchool of Electrical and Electronic Engineering, Wuhan Polytechnic University, Wuhan 430023, ChinaSchool of Resources Environment Science and Technology, Hubei University of Science and Technology, Xianning 437100, ChinaThe existing deep learning-based Airborne Laser Scanning (ALS) point cloud semantic segmentation methods require a large amount of labeled data for training, which is not always feasible in practice. Insufficient training data may lead to over-fitting. To address this issue, we propose a novel Multi-branch Feature Extractor (MFE) and a three-stage transfer learning strategy that conducts pre-training on multi-source ALS data and transfers the model to another dataset with few samples, thereby improving the model’s generalization ability and reducing the need for manual annotation. The proposed MFE is based on a novel multi-branch architecture integrating Neighborhood Embedding Block (NEB) and Point Transformer Block (PTB); it aims to extract heterogeneous features (e.g., geometric features, reflectance features, and internal structural features) by leveraging the parameters contained in ALS point clouds. To address model transfer, a three-stage strategy was developed: (1) A pre-training subtask was employed to pre-train the proposed MFE if the source domain consisted of multi-source ALS data, overcoming parameter differences. (2) A domain adaptation subtask was employed to align cross-domain feature distributions between source and target domains. (3) An incremental learning subtask was proposed for continuous learning of novel categories in the target domain, avoiding catastrophic forgetting. Experiments conducted on the source domain consisted of DALES and Dublin datasets and the target domain consists of ISPRS benchmark dataset. The experimental results show that the proposed method achieved the highest OA of 85.5% and an average F1 score of 74.0% using only 10% training samples, which means the proposed framework can reduce manual annotation by 90% while keeping competitive classification accuracy.https://www.mdpi.com/2072-4292/17/15/2618ALSpoint cloud semantic segmentationdeep learningtransfer learningfew sample learning
spellingShingle Jialin Yuan
Hongchao Ma
Liang Zhang
Jiwei Deng
Wenjun Luo
Ke Liu
Zhan Cai
Transfer Learning Based on Multi-Branch Architecture Feature Extractor for Airborne LiDAR Point Cloud Semantic Segmentation with Few Samples
Remote Sensing
ALS
point cloud semantic segmentation
deep learning
transfer learning
few sample learning
title Transfer Learning Based on Multi-Branch Architecture Feature Extractor for Airborne LiDAR Point Cloud Semantic Segmentation with Few Samples
title_full Transfer Learning Based on Multi-Branch Architecture Feature Extractor for Airborne LiDAR Point Cloud Semantic Segmentation with Few Samples
title_fullStr Transfer Learning Based on Multi-Branch Architecture Feature Extractor for Airborne LiDAR Point Cloud Semantic Segmentation with Few Samples
title_full_unstemmed Transfer Learning Based on Multi-Branch Architecture Feature Extractor for Airborne LiDAR Point Cloud Semantic Segmentation with Few Samples
title_short Transfer Learning Based on Multi-Branch Architecture Feature Extractor for Airborne LiDAR Point Cloud Semantic Segmentation with Few Samples
title_sort transfer learning based on multi branch architecture feature extractor for airborne lidar point cloud semantic segmentation with few samples
topic ALS
point cloud semantic segmentation
deep learning
transfer learning
few sample learning
url https://www.mdpi.com/2072-4292/17/15/2618
work_keys_str_mv AT jialinyuan transferlearningbasedonmultibrancharchitecturefeatureextractorforairbornelidarpointcloudsemanticsegmentationwithfewsamples
AT hongchaoma transferlearningbasedonmultibrancharchitecturefeatureextractorforairbornelidarpointcloudsemanticsegmentationwithfewsamples
AT liangzhang transferlearningbasedonmultibrancharchitecturefeatureextractorforairbornelidarpointcloudsemanticsegmentationwithfewsamples
AT jiweideng transferlearningbasedonmultibrancharchitecturefeatureextractorforairbornelidarpointcloudsemanticsegmentationwithfewsamples
AT wenjunluo transferlearningbasedonmultibrancharchitecturefeatureextractorforairbornelidarpointcloudsemanticsegmentationwithfewsamples
AT keliu transferlearningbasedonmultibrancharchitecturefeatureextractorforairbornelidarpointcloudsemanticsegmentationwithfewsamples
AT zhancai transferlearningbasedonmultibrancharchitecturefeatureextractorforairbornelidarpointcloudsemanticsegmentationwithfewsamples