Efficient Semantic Segmentation for Resource-Constrained Applications with Lightweight Neural Networks

This thesis focuses on developing lightweight semantic segmentation models tailored for resource-constrained applications, effectively balancing accuracy and computational efficiency. It introduces several novel concepts, including knowledge sharing, dense bottleneck, and feature re-usability, which...

Full description

Bibliographic Details
Main Author: Singha, Tanmay
Format: Thesis
Published: Curtin University 2023
Online Access:http://hdl.handle.net/20.500.11937/93644
_version_ 1848765760378241024
author Singha, Tanmay
author_facet Singha, Tanmay
author_sort Singha, Tanmay
building Curtin Institutional Repository
collection Online Access
description This thesis focuses on developing lightweight semantic segmentation models tailored for resource-constrained applications, effectively balancing accuracy and computational efficiency. It introduces several novel concepts, including knowledge sharing, dense bottleneck, and feature re-usability, which enhance the feature hierarchy by capturing fine-grained details, long-range dependencies, and diverse geometrical objects within the scene. To achieve precise object localization and improved semantic representations in real-time environments, the thesis introduces multi-stage feature aggregation, feature scaling, and hybrid-path attention methods.
first_indexed 2025-11-14T11:40:22Z
format Thesis
id curtin-20.500.11937-93644
institution Curtin University Malaysia
institution_category Local University
last_indexed 2025-11-14T11:40:22Z
publishDate 2023
publisher Curtin University
recordtype eprints
repository_type Digital Repository
spelling curtin-20.500.11937-936442023-10-31T01:26:45Z Efficient Semantic Segmentation for Resource-Constrained Applications with Lightweight Neural Networks Singha, Tanmay This thesis focuses on developing lightweight semantic segmentation models tailored for resource-constrained applications, effectively balancing accuracy and computational efficiency. It introduces several novel concepts, including knowledge sharing, dense bottleneck, and feature re-usability, which enhance the feature hierarchy by capturing fine-grained details, long-range dependencies, and diverse geometrical objects within the scene. To achieve precise object localization and improved semantic representations in real-time environments, the thesis introduces multi-stage feature aggregation, feature scaling, and hybrid-path attention methods. 2023 Thesis http://hdl.handle.net/20.500.11937/93644 Curtin University fulltext
spellingShingle Singha, Tanmay
Efficient Semantic Segmentation for Resource-Constrained Applications with Lightweight Neural Networks
title Efficient Semantic Segmentation for Resource-Constrained Applications with Lightweight Neural Networks
title_full Efficient Semantic Segmentation for Resource-Constrained Applications with Lightweight Neural Networks
title_fullStr Efficient Semantic Segmentation for Resource-Constrained Applications with Lightweight Neural Networks
title_full_unstemmed Efficient Semantic Segmentation for Resource-Constrained Applications with Lightweight Neural Networks
title_short Efficient Semantic Segmentation for Resource-Constrained Applications with Lightweight Neural Networks
title_sort efficient semantic segmentation for resource-constrained applications with lightweight neural networks
url http://hdl.handle.net/20.500.11937/93644