A5044517527- Urban Transport and Accessibility
- Face recognition and analysis
- Human Mobility and Location-Based Analysis
- Remote-Sensing Image Classification
- Domain Adaptation and Few-Shot Learning
- Advanced Image and Video Retrieval Techniques
- Face and Expression Recognition
- Video Surveillance and Tracking Methods
- Urban Heat Island Mitigation
- Anomaly Detection Techniques and Applications
- Biometric Identification and Security
- Generative Adversarial Networks and Image Synthesis
- Image Retrieval and Classification Techniques
- Transportation Planning and Optimization
- Automated Road and Building Extraction
- Remote Sensing and Land Use
- Traffic and Road Safety
- Advanced Image Processing Techniques
- User Authentication and Security Systems
- demographic modeling and climate adaptation
- Lymphoma Diagnosis and Treatment
- Innovative Human-Technology Interaction
- Multimedia Communication and Technology
- Myasthenia Gravis and Thymoma
- Digital Mental Health Interventions
The University of Melbourne
2002-2024
ORCID
2022
Faculty of Design
2021
University of Moratuwa
2015-2017
Parks Victoria
2009
Commonwealth Scientific and Industrial Research Organisation
2002
The Royal Melbourne Hospital
1999
Kowloon Hospital
1989-1994
Prince of Wales Hospital
1990
Extracting emotions from physiological signals has become popular over the past decade. Recent advancements in wearable smart devices have enabled capturing continuously and unobtrusively. However, signal readings different wearables are lossy due to user activities, making it difficult develop robust models for emotion recognition. Also, limited availability of data labels is an inherent challenge developing machine learning techniques classification. This paper presents a novel...
This paper presents a summary of the Masked Face Recognition Competitions (MFR) held within 2021 International Joint Conference on Biometrics (IJCB 2021). The competition attracted total 10 participating teams with valid submissions. affiliations these are diverse and associated academia industry in nine different countries. These successfully submitted 18 solutions. is designed to motivate solutions aiming at enhancing face recognition accuracy masked faces. Moreover, considered...
Malware detection plays a crucial role in cyber-security with the increase malware growth and advancements cyber-attacks. Previously unseen which is not determined by security vendors are often used these attacks it becoming inevitable to find solution that can self-learn from unlabeled sample data. This paper presents SHERLOCK, self-supervision based deep learning model detect on Vision Transformer (ViT) architecture. SHERLOCK novel method learns unique features differentiate benign...
Diffusion models have emerged as state-of-the-art generative AI in computer vision, excelling generating high-fidelity and diverse images. These surpass previous architectures quality, generalizability, stability. However, their potential remains largely untapped water resource applications, including flood mapping.This research introduces a diffusion model-based super-resolution approach to upscale coarse-grid hydrodynamic model outputs fine-grid accuracy computationally efficient manner....
An anatomical study of the thoracodorsal arterial system was performed; it focused on angular branch. The aim to document variations this pedicle and delineate area supply inferior angle scapula with a view free bone transfer. A total 81 cadaver dissections were they revealed constant presence artery four vascular patterns origin Selective India ink perfusion studies performed 11 sides in six fresh cadavers demonstrated reliable extent 6 cm vertebral margin 3 lateral scapula. Histologic...
Automatically converting text descriptions into images using transformer architectures has recently received considerable attention.Such advances have implications for many applied design disciplines across fashion, art, architecture, urban planning, landscape and the future tools available to such disciplines.However, a detailed analysis capturing capabilities of models, specifically with focus on built environment, not been performed date.In this work, we investigate biases text-to-image...
The lymphocytes that accompany thymomas express an immature T-cell phenotype, as usually demonstrated by CD1 or TdT immunoreactivity. Even when metastasize occur in ectopic sites, the infiltrating T show this unique contrasting with thymic and nonthymic carcinomas, which typically a mature phenotype (CD1 negative). Therefore, presence of population epithelial tumor strongly supports diagnosis thymoma. availability antibody consistently marks T-cells routine paraffin sections would be great...
Due to the costly nature of remote sensing image labeling and large volume available unlabeled imagery, self-supervised methods that can learn feature representations without manual annotation have received great attention. While prior works explored learning in tasks, pretext tasks based on local-global view alignment remain underexplored. Inspired by DINO, which employs an effective representation structure with knowledge distillation global-local alignment, we formulate two for use...
Extensive research has been conducted exploring associations between built environment characteristics and biking. However, these approaches have often lacked the ability to understand interactions of environment, population bicycle ridership. To overcome limitations, this study aimed develop novel urban biking typologies using unsupervised machine learning methods. We a retrospective analysis travel surveys, infrastructure land use in Greater Melbourne region, Australia. typology, we used...
In recent decades, wildfires have caused tremendous property losses, fatalities, and extensive damage to forest ecosystems. Inspired by the abundance of publicly available remote sensing projects burgeoning development deep learning in computer vision, our research focuses on assessing fire risk using imagery.In this work, we propose a novel dataset, FireRisk, consisting 7 classes with total 91872 labelled images for assessment. This dataset is supplied Wildfire Hazard Potential (WHP) raster...
Estimation of bicycling volumes is essential for the strategic implementation infrastructure and related transport elements policies. Link-level volume estimation models (models that estimate on individual street segments) allow understanding variation in across an entire network at higher spatial resolution than area-level models. Such assist planners to efficiently monitor usage, identify opportunities enhance safety evaluate impact policy interventions. However, given sparsity scarcity...
Many different image stitching algorithms, and mechanisms to assess their quality have been proposed by research groups in the past decade. However, a comparison across algorithms evaluation has not performed before. Our objective is recognize best algorithm for panoramic stitching. We measure robustness of means assessing set panoramas. For itself varied assessment criteria are used, over large range images captured using differing cameras. In an ideal algorithm, resulting stitched should...
Numerous studies have explored associations between bicycle network characteristics and ridership. However, the majority of these been conducted in inner metropolitan regions as such, there is limited knowledge on how various networks relate to trips within across entire regions, size composition study impact association We a retrospective analysis household travel survey data infrastructure Greater Melbourne region, Australia. Seven metrics were calculated (length network, betweenness...
Neural networks are powerful function approximators, yet their ``black-box" nature often renders them opaque and difficult to interpret. While many post-hoc explanation methods exist, they typically fail capture the underlying reasoning processes of networks. A truly interpretable neural network would be trained similarly conventional models using techniques such as backpropagation, but additionally provide insights into learned input-output relationships. In this work, we introduce concept...
Dual-energy X-ray absorptiometry (DXA) scans are one of the most frequently used imaging techniques for calculating bone mineral density, yet fracture risk using DXA image features is rarely performed. The objective this study was to combine deep neural networks, together with images and patient clinical information, evaluate in a cohort adults at least known fall age-matched healthy controls. entire body as, well as isolated hip, forearm, spine (1488 total), were obtained from 478 fallers...
The COVID-19 pandemic has drastically changed accepted norms globally. Within the past year, masks have been used as a public health response to limit spread of virus. This sudden change rendered many face recognition based access control, authentication and surveillance systems ineffective. Official documents such passports, driving license national identity cards are enrolled with fully uncovered images. However, in current global situation, matching should be able match these reference...
This paper describes the experimental framework and results of IJCB 2022 Mobile Behavioral Biometrics Competition (MobileB2C). The aim MobileB2C is bench-marking mobile user authentication systems based on behavioral biometric traits transparently acquired by devices during ordinary Human-Computer Interaction (HCI), using a novel public database, BehavePassDB <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup>...
The recent renaissance in generative models, driven primarily by the advent of diffusion models and iterative improvement GAN methods, has enabled many creative applications. However, each advancement is also accompanied a rise potential for misuse. In arena deepfake generation, this key societal issue. particular, ability to modify segments videos using such techniques creates new paradigm deepfakes which are mostly real altered slightly distort truth. This been under-explored current...