Vulnerable Road Users Detection using Convolutional Deep Feedforward Network

A new convolutional deep feedforward network (C-DFN) is proposed to detect vulnerable road users at 57.9% misclassification rate using Caltech Dataset. Instead of going deeper, three C-DFN is stacked to achieve 43.4% misclassification rate. Part-based C-DFN further reduces the rate of 42.5% to tackl...

Full description

Bibliographic Details
Main Author: Lau, Mian Mian
Format: Thesis
Published: Curtin University 2021
Online Access:http://hdl.handle.net/20.500.11937/83745
_version_ 1848764603442397184
author Lau, Mian Mian
author_facet Lau, Mian Mian
author_sort Lau, Mian Mian
building Curtin Institutional Repository
collection Online Access
description A new convolutional deep feedforward network (C-DFN) is proposed to detect vulnerable road users at 57.9% misclassification rate using Caltech Dataset. Instead of going deeper, three C-DFN is stacked to achieve 43.4% misclassification rate. Part-based C-DFN further reduces the rate of 42.5% to tackle occlusion problem. In addition, investigation of adaptive activation functions are performed to understand the effect of saturated and non-saturated functions in mitigating the vanishing and exploding gradient issues of neural networks.
first_indexed 2025-11-14T11:21:59Z
format Thesis
id curtin-20.500.11937-83745
institution Curtin University Malaysia
institution_category Local University
last_indexed 2025-11-14T11:21:59Z
publishDate 2021
publisher Curtin University
recordtype eprints
repository_type Digital Repository
spelling curtin-20.500.11937-837452021-05-24T07:07:31Z Vulnerable Road Users Detection using Convolutional Deep Feedforward Network Lau, Mian Mian A new convolutional deep feedforward network (C-DFN) is proposed to detect vulnerable road users at 57.9% misclassification rate using Caltech Dataset. Instead of going deeper, three C-DFN is stacked to achieve 43.4% misclassification rate. Part-based C-DFN further reduces the rate of 42.5% to tackle occlusion problem. In addition, investigation of adaptive activation functions are performed to understand the effect of saturated and non-saturated functions in mitigating the vanishing and exploding gradient issues of neural networks. 2021 Thesis http://hdl.handle.net/20.500.11937/83745 Curtin University fulltext
spellingShingle Lau, Mian Mian
Vulnerable Road Users Detection using Convolutional Deep Feedforward Network
title Vulnerable Road Users Detection using Convolutional Deep Feedforward Network
title_full Vulnerable Road Users Detection using Convolutional Deep Feedforward Network
title_fullStr Vulnerable Road Users Detection using Convolutional Deep Feedforward Network
title_full_unstemmed Vulnerable Road Users Detection using Convolutional Deep Feedforward Network
title_short Vulnerable Road Users Detection using Convolutional Deep Feedforward Network
title_sort vulnerable road users detection using convolutional deep feedforward network
url http://hdl.handle.net/20.500.11937/83745