Research Bank
Not a member yet
5195 research outputs found
Sort by
Early career researchers succeeding under a changing research system
No full textThe Royal Society Te Apārangi Early Career Researcher (ECR) forum represents researchers in Aotearoa New Zealand across various research sectors encompassing Crown Research Institutes (CRIs), Industry Training Organisations (ITOs), and universities to better support ECR career development. Despite recommendations to integrate these research sectors, the system is still segmented with rising numbers of PhD graduates, limited post-doctoral opportunities, and challenges associated with a changing research system. Recent efforts, including funding reforms and new fellowship schemes, aim to address these issues but remain insufficient. This paper highlights ongoing disparities and the need for a framework that fosters ECR mobility and professional growth. It calls for strategic reforms in training and funding systems, supporting integrated pathways, equitable opportunities, and fostering of ECRs across diverse research environments
Evaluating deepfake detection models: A comprehensive framework for comparison across diverse datasets
No full textRapid advancements in deepfake technology pose serious threats to public confidence in digital material, cybersecurity, and the ability to authenticate media. Deepfake movies have gotten progressively realistic with new artificial intelligence developments, especially in Generative ANs and autoencoders, which makes detection more difficult. By creating and assessing ML models for recognizing deep fakes and using several datasets to improve generalizability and robustness, our work seeks to solve these difficulties.
Celeb-DF V2, FaceForensics++, CIPLib, Deepfake Detection Challenge Dataset, as well as Fake Celeb AV-V1.2 are among the datasets this work uses. Every dataset is chosen depending on their variance in video alteration techniques, resolution changes, and source authenticity to guarantee a thorough assessment of machine learning models among several deepfake generating approaches. Steps in data preparation augmentation, normalisation, feature extraction are used to improve model training's efficacy.
Then we investigate several deepfake detection models: Forensics Convolutional Neural Networks, Vision Transformers, Masked Autoencoders, Two-Stream Neural Network as well as Common Fake Feature Network in a comparative manner. Every model is evaluated in relation to computing efficiency, generalizing across datasets, and efficacy in reducing detection mistakes. Performance data including validation and training accuracy, loss functions, and error rates including equal error rate, half total error rate, false acceptance rate, and false rejection rate is examined using a standard assessment framework.
While Vision Transformers and Two-Stream Neural Network models show great detection ability, experimental results show that the Common Fake Feature Network model routinely beats others, obtaining almost zero error rates and up to 100% validation accuracy on demanding datasets including Fake Celeb AV-V1.2 and Deepfake Detection Challenge Dataset. Common Fake Feature Network exceptional success can be ascribed to its capacity to use sophisticated feature extraction and classification approaches to capture fine-grained alterations in deepfake films.
The results of this work underline the need of constantly developing machine learning methods to properly fight digital disinformation since they show the need of managing detection accuracy with error minimizing. Furthermore, underlined by this study is the crucial need of strong deep fake identification systems in digital integrity, cybersecurity, and media forensics. This work adds important new perspectives on the continuous battle against deepfake dangers by building a disciplined evaluation approach and contrasting several SotA detecting techniques. Future research will concentrate on increasing model adaptability, computing efficiency, and integration of present time deep fake detection for pragmatic application in digital security systems
Optimising YOLO and ByteTrack for robust vehicle counting and classification in adverse weather: A computer-vision-based traffic monitoring study using New Zealand data
No full textRESEARCH QUESTIONS
• How computer vision and machine learning techniques can be used to develop a robust system that can collect traffic data?
• How will it perform in a real-time scenario?
• How well will this method perform in rain and fog where the visibility is low and in heavily congested traffic?
• Can we capture a New Zealand Dataset representing all weather and traffic conditions?
• Is it going to be challenging to prepare a ground truth from the newly captured video data?
ABSTRACT
Traffic surveillance is critical in modern transportation management systems. It facilitates efficient traffic flow, ensures road safety, and enables informed decision-making. A significant part of traffic surveillance is collecting accurate data on traffic volume and vehicle class. This facilitates adequate and confident decision-making when it comes to future planning.
Conventional traffic volume and class data collection methods, such as loop sensors and pneumatic tubes, have been extensively utilised over an extended period. However, they are less reliable in congested traffic situations as they mostly rely on the length of the vehicle to determine the class of the vehicle. They are also expensive to install or maintain and disrupt traffic in the process. On the other hand, computer vision and deep learning-based approaches have gained significant popularity in recent years and have already been successfully applied in various traffic-related operations. These methods can collect in-depth traffic data from multiple lanes and directions with just one camera, reducing costs significantly while improving data quality.
Computer-vision-based traffic monitoring systems are good at collecting traffic vol ume and class data in general. However, there is room for improvement in extreme weather conditions like heavy rain and heavily congested traffic like stop-and-go sit uations and in classifying complex vehicles like long freight trucks and vehicles with trailers. A computer vision-based traffic monitoring system capable of surpassing tra ditional methods in terms of accuracy and reliability, particularly for measuring traffic volume and class data under challenging weather and traffic conditions, remains an area of ongoing research and development.
This study thoroughly investigates different computer vision models for traffic count and vehicle classification. It proposes a new method to improve the accuracy and detection speed in various adverse weather, lighting, and traffic congestion condi tions aimed at a reliable live monitoring system. This thesis also collects video footage of unique traffic and weather conditions from the New Zealand State Highway to eval uate the proposed method. A total of thirty-five videos covering a total length of over one hour of footage include 8 unique traffic and weather conditions and 7343 vehicles. With these video data, different sizes or sub-versions of You Only Look Once or YOLO (versions 8, 9, 10, and 11) have been thoroughly evaluated against Detection Trans former (DETR), and Faster Region Convolutional Neural Network (Faster-RCNN). A YOLOv10l and a ByteTrack model have been optimised to perform best on both accuracy and detection speed, and the final method is proposed with the optimised version of YOLOv10l and ByteTrack. Out of the box, YOLO and OpenCV-based models gave very inconsistent results where the accuracy fluctuates from 22.19% undercount to 258.97% overcount of the real vehicle count. In this study, multiple weaknesses in the Object-Detection Model and Tracker Model were identified. The proposed method resolved these issues using an optimised YOLOv10l and an optimised ByteTrack model. Overall, it achieved 98.01% accuracy on the total vehicle count, 97.32% on vehicle classification, and 97.67% on accurate lane detection with 17.84 frames per second which is a 32.13% improvement in counting accuracy and 12.64 frames per second faster compared to the out-of the-box YOLO and Open-CV based method. Compared to other versions of YOLO the proposed method outperforms version 8, 9, and 11 by 1.29%, 1.64%, and 0.63% respectively in overall accuracy while maintaining negligible difference in processing time. When it comes to DETR and Faster-RCNN, the proposed method was signif icant faster in processing time and significantly more accurate that Faster-RCNN. It achieved 1.09% higher accuracy with 17.16 FPS faster speed over DETR, and 24.31% higher accuracy with 17.36 FPS faster speed over Faster-RCNN
An ecosystem for the sentient object - Two new taxonomic models for New Zealand museums
No full textIf it could be demonstrated that both indigenous and western artefacts possess a life essence, what kind of ecosystem would they live in, and could this premise be applied to better care for our collections in the wider museum context?
In my thesis, I explore what can be learnt from the cultural and spiritual traditions of indigenous artefacts, (taoka), and how we can adapt this ontological approach to western objects so that we may better understand and care for all our collections. From this, a new taxonomy, ordering of collections, can be formed - one that breaks away from colonial museum conventions.
I propose a new museum model that reorders the grouping and storing of collections by prioritising object-genealogies over museum nomenclature. The model realigns traditional museum systems to better fit and understand the needs of the cultural artefacts and objects in our care, allowing us to consider how object-genealogies can lead to a more logical and culturally appropriate approach to storing museum collections.
The approach has demanded a qualitative methodology, achieved by introducing researched indigenous principles around ‘living’ artefacts, which are then paralleled with western object- narrative case studies to prove the existence of object-ontologies. By prioritising object- genealogies over type, a taxonomy that takes into account all objects as living beings can be created.
This new taxonomy respects indigenous traditions and ensures that all things (humans, taoka, objects) remain connected. While taoka have always carried a life-essence, the reframing of western objects in this perspective provides a way forward toward a fresh perspective for any museum looking for a less colonial approach to collection-care
Places in time: Public space practices for dynamic being in rural China
No full textRESEARCH QUESTION
How can architectural interventions create sustainable public cultural spaces in rural China to enhance the dynamic being between people and architecture?
ABSTRACT
Amid China’s rural revitalization campaign, this project explores adaptive reuse strategies for abandoned buildings and idle spaces in ShuangZha Village, Tianjin. The interventions include a historic water tower, a former village clinic, a vacant lot formed from a filled sewage pond, and mobile ritual spaces modified from trucks—each responding to rural transformation across both material and spiritual dimensions.
While improvements in infrastructure and living standards have rendered some original facilities obsolete, challenges such as cultural loss, population aging, and spatial hollowing persist. Guided by the existential framework of Dynamic Being, the project takes “water”— its memory, use, sharing, and flow—as a narrative thread to reweave connections among people, place, and ritual. Through historical research, field surveys, and spatial interventions, it proposes a design strategy that both restores physical environments and revitalizes ecological perception and collective identity.
Ultimately, the project outlines a rural cultural regeneration model rooted in locality and attuned to change. Drawing on Heidegger's thinking, architecture is framed not as a tool of control but as a restrained, embedded practice—an act of care that acknowledges fragility and expresses love for both people and land. It asks how spatial practice can quietly sustain the continuity of rural life—not only by repairing what is seen, but by nurturing what is felt, remembered, and shared.
GLOSSARY
1. 乡愁 (Xiāng Chóu) – "Nostalgia" The profound longing for one's homeland or past.
2. 小站稻 (Xiǎo Zhàn Dào) – "Xiaozhan Rice" A premium variety of glutinous rice known for its fragrant and tender quality.
3. 炕 (Kàng) – "Heated Brick Bed" A traditional Chinese heated sleeping platform that provides warmth in winter.
4. 囍 (Xǐ) – "Double Happiness" A traditional symbol representing marital bliss and celebration.
5. 福 (Fú) – "Blessing/Good Fortune" A cultural symbol denoting auspiciousness and prosperity.
6. 摊场 (Tān Chǎng) – "Threshing Floor" An open area for grain threshing and sun-drying.
7. 上善若水 (Shàng Shàn Ruò Shuǐ) – "The Highest Virtue is Like Water" A philosophical concept suggesting that supreme goodness flows naturally like water, nourishing without contention.
8. 饮水思源 (Yǐn Shuǐ Sī Yuán) – "When Drinking Water, Remember Its Source" An idiom emphasizing gratitude and remembering one's roots when benefiting from others' contributions.
9. 水到渠成 (Shuǐ Dào Qú Chéng) – "When Water Flows, a Channel Forms Naturally" A proverb meaning success comes naturally when conditions are ripe.
10. 逝者如斯 (Shì Zhě Rú Sī) – "All Things Pass Like the Flowing River" A philosophical observation about the transient nature of time, encouraging appreciation of the present moment
Perception of Auckland Council’s biosecurity Commitment Form by recreational boat users in Tīkapa Moana/Hauraki Gulf
No full textRecreational boating is a pathway for the introduction and reintroduction of invasive species to the islands of Hauraki Gulf/Tīkapa Moana. Potential invasive organisms include, but are not limited to rodents, skinks, ants, and plant pathogens. ‘Check, Clean, Close’ is the primary messaging of the Protect Our Hauraki Gulf campaign by Auckland Council with their Commitment Form acting as a checklist for boat users to self-certify they have performed these three biosecurity actions before heading out to sea. The aim of this study was to assess selected recreational boat users’ comprehension and perception of the Commitment Form in the current biosecurity awareness campaign and identify motivators for undertaking biosecurity actions. An anonymous survey of Hauraki Gulf recreational vessel users was conducted at Auckland boat shows in March and May 2025, as well as at Auckland boat ramps. Questions focused on comprehension and motivators for behaviour change, as well as potential improvements, and the best placement of the form to reach boat users. Of the 357 respondents, 82% agreed or agreed strongly that the commitment form would motivate them to complete preventative biosecurity actions. The motivations behind the desire to act were identified as a. Reducing cost of vessel maintenance (40%) and b. Protecting the marine and island environment (30%). These findings are vital for informing strategies and decision-making to support Auckland Council in their biosecurity messaging for the future protection of the Hauraki Gulf island
Vishing detection over call transcripts using Zero-shot Learning and machine learning
No full textRESEARCH QUESTIONS
• How can Zero-Shot Learning be utilized to detect fraud data without relying on task-specific labeled data?
• How does ZSL tackle the issue of limited vishing data for training models?
• How does ZSL perform compared to traditional ML and DL models?
ABSTRACT
Vishing assaults affect people and organizations by tricking victims into divulging personal information over the phone. Although deep learning models have been used in many fields, creating efficient models is difficult due to complicated datasets and insufficient data. Accuracy in voice-based phishing detection is frequently limited by additional challenges arising from data volume, accessibility, and privacy considerations. These issues are addressed by transfer learning, especially Zero-Shot Learning (ZSL), which uses semantic descriptions and pre-trained models to identify invisible categories.Financial loss, harm to one’s reputation, and data breaches can be avoided by being aware of social engineering techniques like mimicry, urgency, and authority.
In this thesis, a Zero-Shot Learning (ZSL) approach for identifying fraudulent vishing attempts from text-based voice conversation transcripts is proposed. In order to forecast fraud using attribute scores, the model takes textual elements and associates them with social engineering attributes, such as impersonation, urgency, authority, and sensitive information. ZSL is compared with traditional ML models (Naive Bayes, SVM, Logistic Regression, Decision Tree, Random Forest) and LSTM networks in experiments conducted on datasets with 1,000–5,000 entries. Without task-specific labeled data, the ZSL model shows its capacity to generalize and detect fraudulent material with high accuracy between 0.97 and 0.98 across all dataset sizes. By analyzing semantic context, ZSL also successfully manages ambiguous or invisible cases, demonstrating its promise as a scalable, label-efficient method for proactive fraud detection.
The evaluation underscores the ability of Zero-Shot Learning (ZSL) to identify deceptive content by interpreting the semantic context of a message rather than relying on previously labeled data. In one case, the ZSL model classified a message flagged as fraudulent in the dataset as normal due to its low alignment with predefined social engineering traits and low score of tactics, so this instance illustrates how ZSL uses the contextual information of the text compared to traditional models. This shows the ZSL’s capacity to handle unfamiliar or borderline cases with contextual awareness, addressing the research objective of exploring effective fraud detection without task-specific training data
Incremental learning framework for queen bee detection from acoustic beehive signals
No full textRESEARCH QUESTIONS
1. Compared to conventional methods, how does LDA-based feature extraction identify the most important features in a queen bee detection system?
2. In what ways can QR be applied to LDA-transformed features to improve computational efficiency and support scalable real-time queen bee detection?
3. What are the main drawbacks of traditional batch learning methods, and how can incremental learning improve the adaptability of queen bee detection?
4. How does the proposed LDA, QR, and incremental learning scheme perform in terms of memory usage, classification accuracy, and model update speed compared to other machine learning pipelines?
5. Does the reduction of feature dimensionality through LDA feature selection and QR help prevent overfitting and thus improve generalisation in dynamic queen bee detection scenarios?
ABSTRACT
Monitoring queen bee status is the foundation of modern beekeeping. Traditional methods are often disruptive, time-consuming, and unsuitable for the computational demands of processing large-scale acoustic data. Due to these challenges, this research proposed an innovative incremental learning framework that combined:
Linear Discriminant Analysis (LDA) for discriminative feature selection, QR Decomposition for efficient and numerically stable model updates, and Support Vector Machine (SVM) for robust classification. The system processed continuous hive audio streams using the incremental QR-based LDA method, enabling the feature space and classifier to adapt to new data without complete model retraining. The model demonstrated exceptional performance and efficiency when evaluated on three distinct and diverse public datasets (NU-Hive, SBCM, and UrBAN). It consistently maintained high classification accuracy while reducing model update time. A key finding of this study is that the framework exhibits significant advantages in mitigating overfitting. The inherent information abstraction capability during the incremental QR update process enables it to generalize to unseen test data better. Finally, this study validates a scalable, efficient, and accurate real-time audio-based queen bee detection solution. The solution balanced classification performance and dynamic acoustic environment requirements, thereby representing a breakthrough in precision beekeeping technology
Exploring service delivery gaps in childhood otitis media: A focus on Māori and Pacific tamariki (children) in Aotearoa
No full textOtitis media, commonly referred to as "glue ear," is a condition characterised by the presence of fluid in the middle ear, which can manifest in two primary forms. Acute otitis media (AOM) occurs when the fluid buildup is associated with symptoms of acute illness, such as ear pain, fever, upper respiratory tract infection, tinnitus, or vertigo (Auckland Regional Community Pathways, 2025). In contrast, otitis media with effusion (OME) is the presence of fluid in the middle ear without evidence of acute inflammation (Auckland Regional Community Pathways, 2025)
Fist-bumps for men and not for women? When small gestures matter for inclusivity in construction: An Aotearoa New Zealand case study
No full textCreating an inclusive workplace in construction is essential if gender diversity is to be achieved in the sector. Women continue to be underrepresented in the industry, and despite recent initiatives to increase their numbers, attracting and retaining women remains a constant challenge. The discriminatory work environment is the leading deterrent for women considering a career in construction. This research investigates existing barriers and enablers for women in the construction industry, and strategies that can contribute to creating inclusive environments. An explorative case study was conducted on a Tāmaki Makaurau Auckland-based construction company using in-depth semi-structured interviews and questionnaires with six female participants. Key barriers identified were women experiencing difficulty gaining respect from male colleagues, the expectation of physical strength for labour-intensive jobs, and social isolation for women on the job due to pre-existing informal male networks. Participants experienced unintentional or intentional everyday gestures at work that made them question their sense of belonging on construction sites and hindered workplace inclusivity. The findings suggest that a concerted effort must be made to shift the outdated mindset that construction is not a place for women. Furthermore, leaders should implement inclusivity initiatives and recognition for women when merited, and act as role models for others in the sector. Training modules and development programmes on equity and diversity need to be developed and promoted, and employee participation should be mandatory