TransUNet for Cross-Domain Semantic Segmentation of Urban Scenery

Wei Yuen Teh; Ian K. T. Tan

doi:10.1109/ispacs57703.2022.10082854

TransUNet for Cross-Domain Semantic Segmentation of Urban Scenery

Wei Yuen Teh, Ian K. T. Tan

School of Mathematical & Computer Sciences

Wau Labs Sdn Bhd

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

TransUNet is a hybrid architecture that combines a transformer-based encoder with a CNN-based UNet. Originally introduced for semantic segmentation of medical images, we show in our work that TransUNet can be successfully applied to urban scenery datasets commonly used for developing autonomous driving systems. We also explore the performance characteristics of training on multi-domain data from the real world and a simulator, and show that using simulated images to augment a live dataset can improve segmentation performance. Code will be made available at https://github.com/weiyuen.

Original language	English
Title of host publication	2022 International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS)
Publisher	IEEE
ISBN (Electronic)	9798350332421
DOIs	https://doi.org/10.1109/ispacs57703.2022.10082854
Publication status	Published - 31 Mar 2023

Keywords

Autonomous Driving
Domain Adaptation
Semantic Segmentation
Urban Scenery
Vision Transformer

ASJC Scopus subject areas

Artificial Intelligence
Computer Networks and Communications
Computer Science Applications
Computer Vision and Pattern Recognition
Signal Processing

Access to Document

10.1109/ispacs57703.2022.10082854

Cite this

@inproceedings{d35504e2d9e9408dae2250a23dd4677a,

title = "TransUNet for Cross-Domain Semantic Segmentation of Urban Scenery",

abstract = "TransUNet is a hybrid architecture that combines a transformer-based encoder with a CNN-based UNet. Originally introduced for semantic segmentation of medical images, we show in our work that TransUNet can be successfully applied to urban scenery datasets commonly used for developing autonomous driving systems. We also explore the performance characteristics of training on multi-domain data from the real world and a simulator, and show that using simulated images to augment a live dataset can improve segmentation performance. Code will be made available at https://github.com/weiyuen.",

keywords = "Autonomous Driving, Domain Adaptation, Semantic Segmentation, Urban Scenery, Vision Transformer",

author = "Teh, {Wei Yuen} and Tan, {Ian K. T.}",

year = "2023",

month = mar,

day = "31",

doi = "10.1109/ispacs57703.2022.10082854",

language = "English",

booktitle = "2022 International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS)",

publisher = "IEEE",

address = "United States",

}

TY - GEN

T1 - TransUNet for Cross-Domain Semantic Segmentation of Urban Scenery

AU - Teh, Wei Yuen

AU - Tan, Ian K. T.

PY - 2023/3/31

Y1 - 2023/3/31

N2 - TransUNet is a hybrid architecture that combines a transformer-based encoder with a CNN-based UNet. Originally introduced for semantic segmentation of medical images, we show in our work that TransUNet can be successfully applied to urban scenery datasets commonly used for developing autonomous driving systems. We also explore the performance characteristics of training on multi-domain data from the real world and a simulator, and show that using simulated images to augment a live dataset can improve segmentation performance. Code will be made available at https://github.com/weiyuen.

AB - TransUNet is a hybrid architecture that combines a transformer-based encoder with a CNN-based UNet. Originally introduced for semantic segmentation of medical images, we show in our work that TransUNet can be successfully applied to urban scenery datasets commonly used for developing autonomous driving systems. We also explore the performance characteristics of training on multi-domain data from the real world and a simulator, and show that using simulated images to augment a live dataset can improve segmentation performance. Code will be made available at https://github.com/weiyuen.

KW - Autonomous Driving

KW - Domain Adaptation

KW - Semantic Segmentation

KW - Urban Scenery

KW - Vision Transformer

UR - http://www.scopus.com/inward/record.url?scp=85152274656&partnerID=8YFLogxK

U2 - 10.1109/ispacs57703.2022.10082854

DO - 10.1109/ispacs57703.2022.10082854

M3 - Conference contribution

BT - 2022 International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS)

PB - IEEE

ER -

TransUNet for Cross-Domain Semantic Segmentation of Urban Scenery

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this