Open Theses & Dissertations

Permanent URI for this collection

https://unbscholar.lib.unb.ca/handle/1882/12778

Browse

Now showing 1 - 20 of 2001

An evaluation of the first safer supply program in New Brunswick: The River Stone Recovery Centre
(University of New Brunswick, 2024-11) Thornton, Adrienne; Brunelle, Caroline
In response to the opioid poisoning crisis, safer supply programs (SSPs) have been implemented across Canada, and provide individuals access to prescription medications as an alternative to the illicit drug supply. River Stone Recovery Centre’s injectable opioid agonist treatment program is the only SSP currently operating in New Brunswick. Semi-structured interviews were completed with participants (N = 22), and follow-up interviews (N = 18) were conducted six months later, to assess participant’s experiences and outcomes related to their program involvement. Participants reported positive program experiences and impacts, such as improvements in quality of life, that were attributed to the non-stigmatizing care they were receiving, and to the low-threshold, harm reduction approach of the program. The results from this study demonstrate the positive impact of SSPs in reducing the harms associated with the opioid toxicity crisis, but also highlight areas of unmet needs that can inform harm reduction strategies.
Learning based collaborative task allocation
(University of New Brunswick, 2024-10) Muntaha, Mahjabin; Song, Wei
In the digital era, Mobile Crowdsensing Systems (MCS) utilize mobile and wearable devices for large-scale data collection, forming participatory sensor networks. A primary challenge in MCS is collaborative task allocation, where multiple workers must coordinate to complete tasks. Our approach integrates workers’ social connections, recognizing that those with similar backgrounds collaborate more effectively. We frame this task allocation problem as a graph-based combinatorial optimization task, complicated by spatial, temporal, and social constraints. To ad dress the limitations of traditional heuristics, we propose a Heterogeneous Graph Attention Network-based Double Deep Q-Network (HGDQN). The HGDQN agent autonomously explores and learns to address complex scenarios by capturing nuanced worker-task relationships through subgraphs within the heterogeneous graph. Experimental results show that HGDQN surpasses traditional heuristic methods such as Greedy and ACO methods in scalability, adaptability, and generalization, providing a robust solution for collaborative task allocation in MCS.
It’s all in there, no it isn’t: A Critical Discourse Analysis of the Adolescent Education Programme of India as it relates to gender and sexuality hierarchies
(University of New Brunswick, 2024-10) Singh, Rohina; Rogers, Matt W.B.
This qualitative study takes up Critical Discourse Analysis and thematic document analysis to show how power operates through taken-for-granted schooling policies and practices in India. Specifically, this work addresses power by identifying gaps and inequitable discourses in the Adolescent Education Programme (AEP) regarding gender and sexuality. My data analysis has revealed three salient themes, all of which act as mechanisms of gender inequity: (i) the perpetuation of patriarchal gender norms, (ii) the privileging of heteronormativity, and (iii) biases and discrimination against sexual and gender minorities. Sexuality education geared toward youth in India has the potential to address culture-specific gender norms and attitudes before they crystallize. However, for this to happen, it is vital for Indian youth to engage critically with the discourses sexuality education enacts and perpetuates by critiquing, challenging, and questioning issues of power, marginalization, and representation.
The effects of language use on responses in English/East Asian bilingual individuals: A meta-analysis
(University of New Brunswick, 2024-10) Schriver, Jordan; Perunovic, W.Q. Elaine
According to the linguistic relativity hypothesis, language use can influence the way individuals interact with the world (Whorf, 1956). Compelling evidence for this claim comes from research on bilingual individuals. Specifically, experimental studies in which bilingual individuals have been randomly assigned to communicate in one language or another have shown that language use can affect the way that bilingual individuals think (e.g., Lechuga & Wiebe, 2011), feel (e.g., Alvarado & Jameson, 2011), and behave (e.g., So, 2010). An attempt has not yet been made to integrate these findings, and establish a unified understanding of how the language used by bilingual language users affects their behaviours. Therefore, with this dissertation I addressed this gap by conducting a systematic review and meta-analysis, wherein I attempted to gather all available published and unpublished research up until 2022 in which bilingual individuals proficient in English and an East Asian language were assigned to complete a culturally relevant task in either language, and in which outcomes from the task in each language were compared. Through multilevel meta-analysis including 269 effect sizes drawn from 63 independent samples reflecting the responses of 6693 participants, I found an overall significant effect (g = .27), indicating that when language is manipulated, responses are generally consistent with the cultural norms associated with the language being spoken. The magnitude of this language effect, however, was moderated by the East Asian language being used, and the Task Type under examination. The language effect was not moderated by the Design of the study, Language Acquisition Style, the Country or the City in which the study took place. This meta-analysis offers a synthesis of an existing research program and poses intriguing questions that warrant investigation in future studies.
Compilation-based spatial query processing
(University of New Brunswick, 2024-10) Sahni, Rahul; Ray, Suprio
The proliferation of spatial data applications and rising spatial data volumes demand efficient processing capabilities. Although most relational databases support spatial extensions of SQL, they offer limited scalability. Traditional relational database follows a pull-based model of query processing. This is inefficient for processing large volumes of data. Specialized systems, such as those extending Hadoop and Spark, improve scalability but often lack comprehensive SQL support or suffer from the overheads of the pull-based model. This thesis introduces a distributed spatial query processing system using the Push-based query compilation approach, generating C++/UPC++-based query plans for both single node and distributed execution on a high-performance framework using the Partitioned Global Address Space paradigm. It also proposes two new morsel-driven parallelism algorithms for scalable spatial query execution. Experiments on real-world datasets show significant performance gains over leading systems, including Apache Sedona, Citus - a distributed database based on PostgreSQL, and PostgreSQL in single-node configurations.
The Lock: A play about wrongful convictions
(University of New Brunswick, 2024-10) Leggott, Rosie; Finlay, Triny; Falkenstein, Len
Imagine growing up poor in Canada in the 1950s. Your father has PTSD from the war, your mother struggles with addictions, and your best friend dies in your arms. You are sixteen years old when you are sentenced to life in prison for a murder you didn't commit. This is what happens to Kenneth in The Lock. The Lock is a two-act stage play intermixed with poetry that argues the case for modern ideas of prison abolitionism. Kenneth reflects on his story while suffering from Alzheimer's disease. His broken collection of memories confronts issues of rehabilitation, recidivism, and the human cost of incarceration. The integration of poetic elements within the play serves as a powerful medium for conveying the raw humanity and poignant introspection of those living within the confines of the prison system. The Lock challenges narratives of Canadian exceptionalism, highlighting the nation's history of unjust incarcerations, and shows just how much it takes to break the human spirit.
Development and prototyping of a biologically inspired knee prosthesis
(University of New Brunswick, 2024-10) Lufkin, Jacob Andrew; McGibbon, Chris; Carretero, Juan
A design for a biologically inspired knee prosthesis is presented. The design comes from using spring-driven devices in an antagonistic fashion to control both the position and stiffness of the knee joint. This work presents a proof-of-concept design that seeks to both develop an understanding of how to design and build a usable model and compare that physical model against a simulated model of the knee. The design was made in a pragmatic fashion and sources of error were mitigated during development. The final design following an intensive iterative design project allowed for quasi-static stiffness testing. The model was compared against simulated data. The model showed stiffening behavior and control over the position of the tibial component. It also showed a high amount of correlation to the simulated data. This holds promise for the future of the design, as it is proven to be feasible to build and simulations of optimized geometries can be trusted to yield valid data.
Multi-patch integrodifference models and their eigenvalue problems in spatial ecology
(University of New Brunswick, 2024-10) Beykzadeh, Ali; Watmough, James
In the realm of spatial ecology, we grapple with fundamental questions: How can we design effective nature reserves to safeguard the survival of species? In the context of fisheries, how wide can a fishing zone be without compromising the stability of fish populations? These inquiries have fueled my motivation to delve into the subject matter of this thesis. While we acknowledge that precise answers to such questions remain elusive, I have endeavored to contribute to our understanding of these critical issues. Our journey begins with an exploration of integrodifference equations (IDEs) in spatial ecology in Chapter 1. These mathematical models serve as powerful tools for unraveling the intricate spatial and temporal dynamics of populations characterized by discrete generations and continuous spatial domains. Imagine a population confined to a single isolated patch — a scenario akin to a lake surrounded by hard boundaries. Within this patch, there exists a gap devoid of reproduction, effectively separating the population. Consider, for instance, a protected fishing zone within a lake. Our focus in Chapter 2 lies on understanding the persistence of such populations. We model their life cycles using IDEs and present a method to calculate the maximum allowable gap size that ensures population persistence. The concept of critical patch size takes center stage in Chapter 3. It refers to the minimum favorable area below which a population faces the risk of extinction. Our investigation accounts for the demographic and dispersal traits of individuals, recognizing that these traits may vary across patches. Surprisingly, we find that the smallest critical patch size occurs when individuals exhibit a propensity to leave the patch. Conversely, the largest critical patch size arises when boundaries are more restrictive, limiting the chances of individuals leaving the patch. In the patchy landscape, Chapter 4 introduces an approximation method that simplifies equilibrium population calculations. Our approach involves a form of the redistribution approximation tailored for piecewise continuous kernels. The accuracy of our estimate improves as movement biases near patch boundaries intensify. Key factors influencing our estimate include the growth term’s derivative and the deviation of the equilibrium solution from its average across patches.
Rejecting maternal healthism: A grounded theory
(University of New Brunswick, 2024-10) Baxter, Carolyn; Taylor, Petrea
Many women of childbearing age are living with a higher BMI. Challenges among women living with a higher BMI as they become mothers include experiencing weight stigma and judgment from health care providers. Seven women were interviewed, and a grounded theory methodology is used to understand how women cope with the increased physiological risks and the stigma associated with obesity as they become mothers. Maternal healthism places the responsibility for the health of the baby on maternal behaviours leading to feelings of shame when mothers are unable to meet the expectations of what is “healthy.” Rejecting maternal healthism is a productive response to negative judgment; women living with a higher BMI recognize that the problem is maternal healthism, not their body, and not their ability to be a good mother. Education of health care providers related to rejecting maternal healthism can help to better support women as they become mothers.
Key barriers and career challenges that hinder women from advancing in New Brunswick Information and Communication Technology (ICT) sector
(University of New Brunswick, 2024-10) Abbasi, Katya; Shaikh, Ibrahim
This study aims to understand better the key barriers and challenges that limit women’s participation in senior leadership roles within the Information and Communication Technology (ICT) sector in New Brunswick, including identifying effective response strategies. This study adopted a holistic approach to identify the issue's root cause, consisting of a three-level dimensional analysis. Empirical findings were based on the qualitative analysis of a sample of 27 interviews. Interviews were not limited to women operating in the ICT sector; ecosystem stakeholders, such as Not-for-Profit Organizations (NPOs), public officials, governmental representatives, and the representatives of companies in which those women operate were also interviewed. Significant discrepancies were found regarding the description of the issue by different segments. Emerging findings in the context of the New Brunswick ecosystem, such as “Imposter Syndrome” and “Leaky Boat Syndrome,” were considered key barriers to women’s advancement.
Remembering collective violence: A study of the memories of violence for Indian Sikh and Muslim immigrants in Canada
(University of New Brunswick, 2024-10) Maitreyi, Misha; Holtmann, Cathy
This dissertation examines individual and collective memories of collective violence among Indian Sikh and Muslim immigrants in Canada, focusing on the anti-Sikh violence of 1984 and the anti-Muslim violence of 2002 in India. Drawing on qualitative interviews with 19 participants, the research examines the composition of collective memories, the significance of these memories and their impacts on the daily lives of Sikh and Muslim immigrants in Canada. By integrating concepts of collective memory (Halbwachs, 1992), social memory studies (Olick & Robbins, 1998), revised versions of collective consciousness (Durkheim, 1984[1893]) and collective effervescence (Durkheim, 1965[1912]), and religion as power (Woodhead, 2011; Juergensmeyer, 2010; Friedland, 2001), this study underlines participants’ recollections of mob violence, gendered violence, survival strategies, diverse forms of trauma they have endured and still feel while living in Canada. The findings reveal a common narrative regarding the role of state institutions in orchestrating the violence in both cases. Religious nationalism was deliberately deployed in the 2002 violence against the Muslim minority, moving away from the principles of secularism towards the formation of a Hindu ethnostate. The analysis shows commonalities in the recollection of mob violence and their characteristics such as the use of fire and violence against women. I also highlight how participants’ recollections of similar survival strategies foster collective solidarity and resilience within the communities. Additionally, the participants view the violence in political terms rather than attributing blame to any religion or religious community. Ultimately, by analysing the stories and experiences of violence, I draw parallels with scholarly perspectives on mob violence, shared affect, the political nature of violence, and the instrumentalisation of religious nationalism to demonstrate that while memories can be transient in nature, collective memory persists and reveals significant truths consistent with scholarly explanations. By underscoring the importance of understanding collective memories, I argue that such experiences of collective violence influence perceptions of democracy, minority identities, and human rights as they impact the relationship of immigrants with their countries of origin and residence.
Processing, microstructure, and mechanical properties of additively manufactured IN718 nickel-based superalloy
(University of New Brunswick, 2024-10) Hasani, Navid; Mohammadi, Mohsen
Inconel 718 (IN718) is a widely used superalloy in aerospace applications requiring high strength at elevated temperatures. This thesis comprehensively explores the processing, microstructure, and mechanical behavior of IN718 manufactured by wire arc additive manufacturing (WAAM) and via laser powder bed fusion (LPBF). First, WAAM-produced hybrid IN718-S275 components were investigated, examining interfacial characteristics and texture evolution. Notably, laves phase persistence was observed near the interface even after solution treatment. Neutron diffraction was used to validate the texture of the hybrid components where a strong texture parallel to the build direction in WAAM-IN718 was identified. Elastic-field models were utilized to understand dislocation mobility and Peierls-Nabarro stress, elucidating the role of heat treatment in modifying mechanical properties. Next, differential scanning calorimetry (DSC) was employed to optimize the solutionizing temperature for the complete dissolution of undesirable phases (δ and laves) in LPBF-IN718, with subsequent microstructural characterization. This led to the elimination of micro-segregation and significant Laves dissolution, resulting in a hardness comparable to wrought IN718 alloys. Furthermore, the dynamic deformation behavior of LPBF-IN718 was studied under various elevated strain rates. AMS 5664 heat treatment resulted in a remarkable 28% increase in ultimate compressive strength. Microstructural analysis revealed the presence of strengthening γ" and γ' phases in abundance, and high-density dislocation networks was observed. The influence of strain rate on grain size, texture, and adiabatic shear band formation was thoroughly investigated. The research presented in this thesis provides a comprehensive understanding of how processing techniques and post-fabrication treatments influence the microstructure and mechanical behavior of IN718. This knowledge contributes to optimizing manufacturing processes and developing tailored IN718 heat treatments for aerospace applications. Additionally, this work offers valuable insights into the mechanical response of additively manufactured IN718 under high-strain-rate loading conditions, enhancing the understanding of its performance in critical aerospace and engineering applications.
Privacy-preserving data analytics in advanced metering infrastructure utilizing TEE
(University of New Brunswick, 2024-10) Kariznovi, Arash; Mandal, Kalikinkar
With the rise of the smart grid, modern electrical grids now support two-way communication of energy and data, enabling system optimization through data analytics. However, this also introduces cybersecurity vulnerabilities. While research has focused on using smart meter data to enhance grid performance, security and privacy concerns remain underexplored. This research proposes a secure and privacy-preserving framework for smart meter data transmission and analytics. It combines lightweight cryptography and transport layer security for end-to-end data protection, while Intel SGX ensures private data processing in the cloud. We implemented an efficient LSTM model for energy consumption prediction, demonstrating the framework’s practicality. Our approach balances security, privacy, and functionality, allowing data owners to retain control while leveraging third-party cloud resources.
Achieving secure multicast communication for IoT devices in Aging in Place system
(University of New Brunswick, 2024-10) Gui, Jinkun; Lu, Rongxing
The integration of IoT devices in Aging in Place (AiP) scenarios enhances elderly care but introduces security and privacy challenges. This thesis proposes two schemes to address these issues: an efficient multicast authenticated encryption scheme and a threshold authenticated encryption scheme. The former ensures secure and efficient communication among multiple IoT devices by utilizing Merkle Tree, prefix encoding, XOR filter, and ASCON encryption. The latter uses ElGamal threshold decryption and a binary fuse filter for secure group communication, allowing dynamic membership changes. Both schemes provide robust security, ensuring data confidentiality and integrity while being computationally efficient. These contributions enhance the security and privacy of IoT-enabled AiP systems, benefiting elderly individuals and their caregivers.
Rapid screening and optimization of adsorbents by combining molecular simulation with rapid experimental methods
(University of New Brunswick, 2024-10) Haeri Nejad, Masoud; Eić, Mladen
Adsorption has proved to be an efficient means for removal of pollutants from the atmosphere. In addition to traditional adsorbents, such as silica gel, activated carbon and zeolites, new adsorbents have been introduced which include the metal-organic frameworks (MOF’s). The number of MOF structures is enormous and therefore ranking them for any specific application requires time-saving approaches. In this work, a novel approach dubbed “rapid screening” was introduced. It consists of (1) molecular simulation for prediction of adsorption isotherms followed by (2) the rapid experimental technique of the zero-length-column chromatography (ZLC) for isotherm confirmation. This approach was applied to the example case of screening and modifying metal-organic frameworks (MOFs) adsorbents to remove wasted inhalation anesthetic agents (IAA). IAA adsorption isotherms for three MOFs were predicted using molecular simulation (MS). For successful simulation of these large, branched, and polar adsorbates, an all-atom force field was also developed which accounted for the flexibility of the adsorbate IAA molecules. Using Continuous Fractional Component Monte Carlo (CFCMC) algorithms proved crucial to speed up the simulations. Predictions from MS results were subsequently verified by performing experimental adsorption measurements using traditional methods followed by the faster ZLC technique, with necessary modifications for vapors. ZLC was shown to be able to replace the traditional volumetric or gravimetric adsorption methods with acceptable accuracy. Next, the utility of rapid screening approach was verified for structural modification of a promising adsorbent for enhancing its IAA removal capability. Several structural modifications, including grafting, anion exchange and functionalization of benzene rings were proposed to the alter the pristine structure. New procedures were developed to create of new crystal information files (CIFs) for the modified structures to replace the CIF for the pristine structure as input to molecular simulations. The MS- predicted a two-fold of IAA adsorption capacity for MIL-101-Cr@NH3 (X=F-, OH-) compared with the unmodified structure. Experimental syntheses and rapid ZLC measurements confirmed this prediction. This approach, presented as proof of concept, is applicable to a larger number of materials by procuring or performing syntheses only of those structures that molecular-simulation screening has selected as the most promising adsorbent.
Ideological prejudice among religious fundamentalists and atheists
(University of New Brunswick, 2024-10) McPhee, Rory; Best, Lisa
Numerous religious variables have been linked to a variety of prejudicial attitudes. Much of this research has repurposed measures from the study of ethnic prejudice. Unfortunately, these measures may not always be appropriate as members of religious groups are primarily distinct from each other due to ideological, rather than innate, differences. The goal of this study was to test the possibility that such prejudicial attitudes directed are partially targeted at outgroup teachings, rather than members, by having participants directly evaluate the morality of a sampling of teachings from several religious affiliations, with the authorship of the teachings scrambled to obscure their true sources. Significant relationships were found to influence assessments via either ingroup favoritism or outgroup discrimination. This study offers a versatile and less ambiguous method of assessing prejudice compared to some existing measures. Other novelties included the assessment of atheists and of how personality might play into religious prejudice.
Camera-LiDAR registration using LiDAR feature layers and deep learning
(University of New Brunswick, 2024-10) Leahy, Jennifer; Jabari, Shabnam
This thesis focuses on a new pipeline reducing registration error between optical camera images and LiDAR data, integrating the strengths of both modalities to improve spatial awareness. The first part presents an approach that enhances aerial camera-LiDAR correspondences through weighted and combined LiDAR feature layers comprising intensity, depth, and bearing angle attributes. Correspondences are attained using a 2D-2D Graph Neural Network pipeline and then registered using a 6-parameter affine transformation model, demonstrating pixel-level accuracies that improve its baselines. The second part introduces a new method for camera-LiDAR registration when the modalities come from different projection models, using combined LiDAR feature layers with state-of-the-art deep learning matching algorithms. We evaluate the SuperGlue and LoFTR models on terrestrial datasets from the TX5 scanner, and from a custom-made, low-cost Mobile Mapping System named SLAMM-BOT, across diverse scenes. Registration is achieved using collinearity equations and RANSAC.
Petrogenetic analysis of arc-related Devonian magmatic rocks in relation to the formation of magmatic hydrothermal Cu±Mo±Au deposits in the New Brunswick segment of the Northern Appalachians
(University of New Brunswick, 2024-10) Yousefi, Fazilat; Lentz, David R.; McFarlane, Christopher R. M.
Devonian adakitic porphyritic intrusive rocks in New Brunswick (NB) are associated with porphyry Cu-Mo-Au mineralization and are similar to Cu-Mo-Au porphyries in Québec and Maine, implying a shared genesis. These granitoids are similar petrologically and feature oxidized I-type arc-like compositions with SiO2 ≥66.5 wt.% and Al2O3 >15.5 wt.%, among other distinctive geochemical characteristics. Trace element compositions characterize the link between adakite formation, slab failure, and porphyry Cu-Mo-Au system potential. These rocks are enriched in Cs, Rb, and Ba, but depleted in Nb, Ta, P, and Ti, reflecting subduction-related magmatism. The slab breakoff process involves the breaking of part of the subducted oceanic plate, leading to asthenosphere upwelling and selective partial melting of the slab, the suprasubduction zone lithospheric mantle, the upwelling mantle, and the basal continental crust; transpressional to transtensional tectonics can facilitate magma ascent and emplacement in the post-subduction setting. Zircon, because of its refractory nature and prevalence in igneous rocks, is useful for assessment of magmatic evolution, specifically its oxygen and U-Th-Pb isotopic compositions. This information can be used to predict magmatic fertility in terms of Cu ± Mo ± Au porphyry mineralization. LA-ICP-MS analyses of zircons from the adakitic intrusions investigated reveal diverse compositions, reflecting varied magmatic conditions. The Zr/Hf and Ce/Ce* values provide insights into porphyry fertility, aiding mineralization potential assessment. This study also employs petrography, µXRF-EDS mapping, and LA-ICP-MS to analyze titanite compositions in these oxidized I-type granitoids. Titanite is enriched in high field strength elements (Ta, Zr, Hf, Th, U), rare earth elements, and Sr and can serve as a petrogenetic and metallogenic indicator. Zr-in-titanite thermometry reveals varied crystallization temperatures (784° to 1098°C) among the intrusions investigated. Magnetite phenocrysts in these intrusions have euhedral to subhedral morphology and lack oscillatory zoning (according to SEM-BSE imaging). The diverse trace element compositions offer insight into the magmatic to magmatic-hydrothermal systems. EPMA analysis reveals variable concentrations of SiO2, Al2O3, MgO, Mn, Fe, Ti, Cr, and V in magnetite, with distinguishing features for various deposit systems. When comparing Ti+V with Al+Mn concentrations in magnetite crystals, all samples exhibit similar crystallization temperatures (>500ºC).
Shadow of the Living Brightness
(University of New Brunswick, 2024-09) Béchard, Dominique; Sinclair, Sue
Shadow of the Living Brightness comprises a collection of poems informed by theories of unknowing in Christian mystical theology. Although written from a position of secular faith, the dissertation pulls from the gloomy, passionate writings of the Christian mystics (Gregory of Nyssa, Meister Eckhart, Simone Weil) to further ideas about negative capability, the metaphorical image, desire, loneliness, and suicide. These poems attempt to embody what I value in much unorthodox, mystical theology: a focus on the emotion of religion, and a desire for transcendence that has less to do with overcoming aspects of the physical world, and more to do with uncovering layers of proximity to unknown modes of being. The critical introduction to the poetry is made up of four interconnected sections. I look at the metaphorical image’s unknowable properties, as partially defined by Robert Bly’s theory of the emotive imagination (aka “deep image”) and Federico Garcia Lorca’s theory of duende, and I evaluate the paradox of divine union in mystical theology— mysticism’s insistence on the annihilation of selfhood to be near the incomprehensible divine. I also examine mystical loneliness and the concept of objectless desire (desire with no definable end-consolation). Finally, I look at the possibilities of an unfixed, unknowable despair and how these interact with suicidality. Rather than “remaining content” with not-knowing (as Keats famously asks of a poet), the Christian mystics make a point of embracing the agony of unknowing. Mysticism, based on my readings, allows for the painful sense of being meaningfully torn between the corporeal and the transcendent. It relies on ideas of God that abandon the possibility of consolation: “The extreme greatness of Christianity lies in the fact that it does not seek a supernatural remedy for suffering, but a supernatural use for it" (Weil 132).
Deformation estimation of industrial objects from a single image
(University of New Brunswick, 2024-09) Eivazi Adli, Sahand; Dubay, Rickey; Pickard, Joshua K.; Sun, Grace
Deformations introduced during the manufacturing process of plastic components degrade the accuracy of their 3D geometric information, hindering computer vision-based inspection. This phenomenon is prevalent among the primary plastic products where the objects are devoid of texture. This work proposes a solution for the deformation estimation of texture-less plastic objects using only a single RGB image. This solution encompasses a unique image dataset of five deformed parts, including both real-world and synthetic images, a novel method for generating mesh labels, sequential deformation, and a training model based on graph convolution. The sequential deformation method overcomes the prevalent chamfer distance algorithm in generating precise mesh labels. The model achieves a sub-millimeter accuracy on synthetic images and approximately 2.0 mm on real images, with an average testing time of 1.5 s on the Google Colab’s resources. The model’s high precision and speed make it suitable for real-world applications.

Browse

Recent Submissions