arrow
返回

TransNet: Transformer-Based Point Cloud Sampling Network

delete2023-05-11
delete3
delete
OA
AI
H
Hookyung Lee
J
Jaeseung Jeon
S
Seokjin Hong
J
Jeesu Kim
J
Jinwoo Yoo *
DOI:10.3390/s23104675delete
delete原文链接
delete分享
delete收藏
查看原文
摘要

摘要

En 中文
As interest in point cloud processing has gradually increased in the industry, point cloud sampling techniques have been researched to improve deep learning networks. As many conventional models use point clouds directly, the consideration of computational complexity has become critical for practicality. One of the representative ways to decrease computations is downsampling, which also affects the performance in terms of precision. Existing classic sampling methods have adopted a standardized way regardless of the task-model property in learning. However, this limits the improvement of the point cloud sampling network's performance. That is, the performance of such task-agnostic methods is too low when the sampling ratio is high. Therefore, this paper proposes a novel downsampling model based on the transformer-based point cloud sampling network (TransNet) to efficiently perform downsampling tasks. The proposed TransNet utilizes self-attention and fully connected layers to extract meaningful features from input sequences and perform downsampling. By introducing attention techniques into downsampling, the proposed network can learn about the relationships between point clouds and generate a task-oriented sampling methodology. The proposed TransNet outperforms several state-of-the-art models in terms of accuracy. It has a particular advantage in generating points from sparse data when the sampling ratio is high. We expect that our approach can provide a promising solution for downsampling tasks in various point cloud applications.
Keyword:
deep learning
transformer
self-attention
multi-head attention
point cloud
down sampling
classification
network
AI总结

AI总结

对已上传原文的论文进行重点信息的提取,主要内容包括:简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。

期刊

Sensors 封面图
Sensors
IF:
3.5
论文数:
7.2W
被引数:
20.9W

机构

P
pusan national university
学者数:
2.1W
论文数: 1.9W
被引数: 20
K
kookmin university
学者数:
3.0K
论文数: 3.3K
被引数: 2
引用论文

引用论文

The myeloma cell antigen syndecan‐1 is lost by apoptotic myeloma cells
err2001-12-25
err0
errOAAI
errMichel Jourdan; Martine Ferlin; Eric Legouffe; Mira Horvathova; Janny Liautard; Jean FranÇois Rossi; John Wijdenes; Jean Brochier; Bernard Klein
err分享
err收藏
THE EFFECT OF MESENCHYMAL STEM CELL DERIVED MICROVESICLES IN REPAIR OF FEMORAL CHONDRAL DEFECTS IN DOGS
err2018-12-25
err0
errOAAI
errDina Sabry; Ashraf Shamaa; Mohamed Amer; Omar El-Tookhy; Ahmed Abdallah; Dalia Mohamed Abd El Hassib; Eman Amer; Azza Elamir
err分享
err收藏
Alternative solid state routes to mixed metal oxides (LnCrO3, LnFeO3)
err1996-06-01
err0
PREAI
errI.P. Parkin; A.V. Komarov; Q. Fang
err分享
err收藏
Artificial Neural Networks to Predict the Power Output of a PV Panel
err2014-01-01
err0
errOAAI
errValerio Lo Brano; Giuseppina Ciulla; Mariavittoria Di Falco
err分享
err收藏
Automatic Instrument Segmentation in Robot-Assisted Surgery Using Deep Learning
err
IF0
err2018-03-03
err0
errOAAI
errAlexey A. Shvets; Alexander Rakhlin; Alexandr A. Kalinin; Vladimir I. Iglovikov
err分享
err收藏
Guillain-Barré syndrome following COVID-19: a newly emerging post-infectious complication
err2020-06-14
err0
errOAAI
errSilas Webb; Victoria CJ Wallace; David Martin-Lopez; Mahinda Yogarajah
err分享
err收藏
学者 查看更多内容