UWSpace

UWSpace is the University of Waterloo’s institutional repository for the free, secure, and long-term home of research produced by faculty, students, and staff.

Depositing Theses/Dissertations or Research to UWSpace

Are you a Graduate Student depositing your thesis to UWSpace? See our Thesis Deposit Help and UWSpace Thesis FAQ pages to learn more.

Are you a Faculty or Staff member depositing research to UWSpace? See our Waterloo Research Deposit Help and Self-Archiving pages to learn more.

Photo by Waterloo staff

Communities in UWSpace

Select a community to browse its collections.

Recent Submissions

  • Thumbnail Image
    Item type: Item ,
    Towards Adaptable and Deployable Wildlife Detection from Aerial Imagery
    (University of Waterloo, 2026-04-21) Hsiao, Jayden
    Reliable wildlife monitoring is critical for biodiversity conservation, yet large-scale aerial surveys remain constrained by labour-intensive manual counting and limited model generalization. While camera trap–based approaches are used for species identification and behavioural monitoring at fixed locations, aerial imagery acquired from drones and aircraft enables large-area coverage and is commonly used to estimate population abundance, species distributions, and temporal trends through repeated surveys. However, these workflows remain heavily reliant on manual annotation. Although recent advances in computer vision enable automated wildlife detection, many existing approaches rely on species-specific training data, exhibit poor transferability across environments, and lack mechanisms to incorporate expert corrections into iterative model improvement. These limitations hinder scalability and long-term deployment in operational conservation workflows. We propose OpenWildlife as an open-vocabulary, multi-species wildlife detection framework designed for RGB aerial imagery captured from drones or aircraft, together with a human-in-the-loop (HITL) annotation system that supports incremental model refinement. The framework adapts a language-grounded detection architecture to allow species specification through natural language prompts, enabling flexible identification across terrestrial and marine environments without retraining for each new category. Trained on 15 publicly available wildlife datasets, the model achieves up to 0.981 mAP50 (mean Average Precision at 50% overlap, a standard object detection metric) under fine-tuning and 0.597 mAP50 on seven datasets containing novel species in zero-shot settings, demonstrating generalization to unseen species across diverse vertebrate groups, including mammals and birds. To support practical deployment, the detection model is integrated into a HITL workflow that combines automated pre-labelling, regional human correction, and incremental fine-tuning. While these components are individually well-established, the proposed system integrates them into a unified workflow tailored to aerial wildlife imagery. This design avoids exhaustive full-image annotation in dense scenes and enables expert feedback to directly improve subsequent model iterations. A case study conducted with the Arctic Eider Society on high-resolution aerial surveys of eider ducks in Arctic Canada demonstrates practical impact: the system achieves 77.6% recall with a 22.2% counting error while reducing annotation time by 87.4% compared to fully manual labelling, demonstrating its applicability for semi-automated population abundance estimation from aerial surveys. These results demonstrate that combining open-vocabulary detection with human-in-the-loop learning provides a scalable and adaptable approach for wildlife monitoring, enabling efficient and consistent large-area surveys across diverse species and environments.
  • Thumbnail Image
    Item type: Item ,
    FATal Effect: Investigating causative agents of TDP-43 palmitoylation and mislocalization in ALS
    (University of Waterloo, 2026-04-21) Perry, Cailyn
    Amyotrophic lateral sclerosis (ALS) is a lethal neurodegenerative disease characterized by the progressive loss of upper and lower motor neurons. ALS is highly heterogeneous and has no known singular cause. Of the more than 40 genes associated with ALS, TARDBP stands out, as the encoded protein TDP-43 is known to accumulate in 97% of all ALS cases, regardless of etiology. In these cases, TDP-43 mislocalizes from the nucleus to the cytoplasm resulting in a toxic loss of its native function as an RNA/DNA binding protein, and a toxic gain of cytoplasm accumulation. In neurodegenerative diseases, like ALS, protein aggregation is the most common pathological hallmark. Protein aggregation is often preceded by mislocalization and is an indicator of underlying proteostasis deficiencies. Palmitoylation, the post-translational addition of the 16-carbon fatty acid palmitate, is a reversible lipid modification that, among many things, directs protein membrane trafficking. Palmitoylation is facilitated through the actions of palmitoyl transferases (PATs). Dysregulation of palmitoylation has been linked to proteostasis deficiencies in numerous neurodegenerative diseases, including ALS. Consequently, TDP-43 mislocalization in ALS may be caused by aberrant palmitoylation. This project aims to characterize how TDP-43 localization and palmitoylation are regulated by PATS within the context of ALS. This study confirms that TDP-43 and ALS linked proteins are palmitoylated, and that mutations in TDP-43 result in altered palmitoylation. Understanding how palmitoylation of TDP-43 impacts cellular stress response and recovery within a variety of models will be integral to identifying targets for rescuing proteostasis deficiencies in neurodegenerative disease.
  • Thumbnail Image
    Item type: Item ,
    Pore-Scale Modeling of Platinum Degradation in PEM Fuel Cells
    (University of Waterloo, 2026-04-20) Agravante, Gerard
    Heavy-duty vehicles rely primarily on diesel engines and produce disproportionately higher emissions in the transportation sector. They also lack feasible zero-emission alternatives due to their demanding range and load requirements. Polymer electrolyte membrane (PEM) fuel cells can offer a promising solution due to their long range, low weight, and fast refueling times. However, their application for heavy-duty vehicles is limited by durability issues which limit the fuel cell lifetime. In particular, platinum catalyst degradation in the cathode catalyst layer (CCL) is a critical degradation mechanism. Understanding the mechanisms of platinum degradation through physics-based modeling is essential for developing more durable fuel cells. This thesis first presents an in-depth literature review of physics-based modeling approaches for platinum degradation, classifying the most widely used modeling frameworks and examining how recent studies have utilized and expanded upon them. The review then identified a key gap: the lack of pore-scale models capable of investigating the effects of the CCL microstructure and spatial properties on platinum degradation. To address this gap, a transient pore-scale model of platinum degradation is first developed using pore network modeling. The model couples a performance solver (which simulates the oxygen reduction reaction) with a degradation solver (which simulates the degradation mechanisms of platinum dissolution, oxidation, and mass loss to the membrane). The model is able to reproduce key degradation trends including spatial variations in surface area loss and transient changes in platinum oxide coverage and ion concentration. Importantly, the pore-scale resolution reveals new microstructural insights, such as how constrictions between agglomerates lead to localized degradation and how the length of transport path correlates with platinum loss. Building on the model developed, microstructural and spatial design strategies were systematically investigated. Strategies through modifications in the porosity, agglomerate size, and platinum placement were evaluated for their effects on performance and degradation. It was found that: (1) increasing porosity uniformly or near the membrane reduces platinum loss but worsens performance, although mid-CCL constrictions can minimize this trade-off, (2) reducing agglomerate size enhances performance and has negligible impact on degradation, and (3) increasing platinum loading near the membrane through larger particle sizes promotes more uniform degradation. A combined case study integrating these strategies demonstrates simultaneous improvements in initial performance, total degradation, and degradation uniformity. These results indicate that while individual strategies have distinct trade-offs, combining complementary strategies can retain their advantages while minimizing the drawbacks. Overall, the works presented in this thesis demonstrate the utility of pore-scale modeling for uncovering microstructural insights and identifying mitigation strategies for platinum degradation through the CCL design, which can ultimately contribute to the advancement of PEM fuel cells as a viable clean energy technology for heavy-duty vehicles.
  • Thumbnail Image
    Item type: Item ,
    Numerical Investigation of Infrasound Generated by Clear-Air Turbulence
    (University of Waterloo, 2026-04-20) Drapeau, Christopher
    Clear-air turbulence (CAT) represents a major hazard to the aviation industry because it occurs without visible indicators and cannot be detected by pilots or onboard instruments. This study investigates whether CAT generates infrasonic emissions that could potentially be used for remote detection. High-resolution large-eddy simulations were performed using the Weather Research and Forecasting model to produce atmospheric environments associated with two CAT encounters: a mountain-wave event over Wyoming in 2020 and a shear-driven event over Illinois in 2023. Acoustic source terms were computed from the simulated flow fields using a hybrid acoustic analogy framework to estimate the acoustic pressure at distant observer locations. When acoustic sources were computed using velocity fluctuations relative to the mean flow, the CAT region in the Wyoming case produced acoustic emissions approximately 22--29 dB stronger than the background turbulence, revealing a clear acoustic energy enhancement associated with CAT. However, further investigation incorporating the mean flow and thermodynamic sources demonstrated that the background acoustic field increased substantially due to amplification by the strong terrain-driven mean flow (Wyoming) and underlying convective processes (Illinois). These cases represent particularly energetic atmospheric conditions that reduce the apparent contrast of the CAT signal, producing only modest overall sound pressure level increases relative to the background. Importantly, the thermodynamic sources related to potential temperature remained the principal contribution within the CAT regions. This suggests that the sharp potential temperature gradients and overturning motions play a primary role in CAT-related acoustic emissions, rather than turbulence-generated sound alone. These results demonstrate that CAT can generate measurable acoustic emissions even in realistic and complex atmospheric environments, supporting further investigation of infrasound as a potential remote-detection tool.
  • Thumbnail Image
    Item type: Item ,
    Robust Hardware-Assisted Malware Detection
    (University of Waterloo, 2026-04-20) Propp, Eli
    Malware detection using hardware performance counters (HPCs) offers a promising, low-overhead approach for monitoring program behaviour, as shown in prior work. However, a fundamental architectural constraint, that only a limited number of hardware events can be monitored concurrently, creates a significant bottleneck, leading to detection blind spots. Prior work has primarily focused on optimizing machine learning models for a single, statically chosen event set, or an ensemble of models over the same feature set. We argue that robustness requires diversifying not only the models, but also the underlying feature sets (i.e., the monitored hardware events) in order to capture a broader spectrum of program behaviour. This observation motivates the following research question: Can detection performance be improved by trading temporal granularity for broader coverage, via the strategic scheduling of different feature sets over time? To answer this question, this thesis proposes Hydra, a novel detection mechanism that partitions execution traces into time slices and learns an effective, stochastic schedule of feature sets and corresponding classifiers for deployment. By cycling through complementary feature sets, Hydra mitigates the limitations of a fixed monitoring perspective. Experimental evaluation shows that Hydra significantly outperforms state-of-the-art single-feature-set baselines, achieving at least a 19.32% improvement in F1 score and a 60.23% reduction in false positive rate. These results underscore the importance of feature-set diversity and establish strategic multi-feature-set scheduling as an effective principle for robust, hardware-assisted malware detection.