Lakehead University Knowledge Commons

Knowledge Commons is an open access repository for scholarship and research produced at Lakehead University. It is a free and secure repository for LU faculty, students, staff, and researchers to preserve and present their scholarship.

Communities in KnowledgeCommons

Select a community to browse its collections.

Now showing 1 - 3 of 3

Recent Submissions

Enhancing agricultural semantic segmentation: architectural innovations and strategies for reducing model complexity
(2025) Ling, Chee Mei; Akilan, Thangarajah; Yassine, Abdulsalam; Dzhamal, Amishev; Zhou, Yushi
Agricultural image semantic segmentation plays a vital role in precision agriculture, enabling accurate analysis of visual data to enhance crop management and optimize resource use. However, achieving high segmentation accuracy while maintaining computational efficiency remains challenging, particularly for real-time systems and edge devices. This thesis presents a two-phase research effort toward an efficient and scalable segmentation framework for high-resolution agricultural imagery. In the first phase, an effective model was developed using a novel Dual Atrous Separable Convolution (DAS-Conv) module integrated into a DeepLabV3 backbone. The DAS-Conv module optimizes the balance between dilation rates and padding size to enhance contextual representation without extra computational cost, while a skip connection between encoder and decoder stages improves fine-grained feature recovery. Despite its lightweight design, the model achieved strong results on the Agriculture-Vision benchmark, demonstrating over 66% higher efficiency compared to transformer-based state-of-the-art models. In the second phase, the framework was extended to DAS-SK, which integrates Selective Kernel (SK) attention into the DAS-Conv module to strengthen multi-scale feature learning and adaptability. The enhanced Atrous Spatial Pyramid Pooling (ASPP) module captures both fine local structures and global context, while a dual-backbone design (MobileNetV3-Large and EfficientNet-B3) further improves representation and scalability. Across three benchmark datasets—LandCover.ai, VDD, and PhenoBench—DAS-SK consistently demonstrates superior efficiency–accuracy trade-offs and notable improvements over its predecessor, DAS. On LandCover.ai, DAS-SK achieves 86.25% mIoU, surpassing DAS by +3.17%, using 10.68M parameters and 11.25 GFLOPs. Although it employs slightly more parameters than DAS, the model remains far lighter than hybrid systems such as Ensemble UNet and transformer models like SegFormer MiT-B2. DAS-SK also achieves higher overall efficiency compared with DAS, demonstrating that the added SK attention and dual-backbone design translate directly into improved segmentation quality. A similar trend is observed on the VDD dataset. DAS-SK attains 79.45% mIoU, improving on DAS by +2.25%, while operating with 10.68M parameters and 43.52 GFLOPs. Although the smaller DAS backbone enables slightly higher FPS, DAS-SK delivers the best accuracy–efficiency balance overall, achieving the highest efficiency score of 9.12%, outperforming transformer models whose parameter counts range from 27M to 234M. On the PhenoBench dataset, DAS-SK again provides the highest performance, reaching 85.55% mIoU compared to DAS at 82.53% (+3.02% improvement). The computational cost remains moderate at 45.00 GFLOPs, versus 25.23 GFLOPs for DAS, but the efficiency gain is substantial—10.09% for DAS-SK versus 7.43% for DAS—highlighting the benefits of improved feature selection and multi-scale fusion. Despite introducing a modest computational increase (typically 1.8× more GFLOPs), DAS-SK consistently delivers 2–3% higher mIoU and markedly stronger multi-scale feature discrimination than its ancestor DAS. Combined with parameter counts that remain significantly below those of modern transformer and hybrid models, DAS-SK offers a practical, lightweight, and high-performing solution for real-time agricultural monitoring and remote sensing in resource-limited environments, where accuracy, efficiency, and scalability are equally critical.
Design and optimization of heterogeneous coded distributed computing
(2025) Zhang, Siyu; Deng, Yong; Rathore, M. Mazhar; Akilan, Thangarajah; Zhou, Yushi
The massive increase in data volume in recent years has posed significant challenges for traditional data processing systems. Although distributed computing has been considered as an effective solution, its efficient implementation faces the challenge of the high communication overhead incurred by data exchange (shuffling) between workers. Coded Distributed Computing (CDC) has been proposed by utilizing coded multicasting to reduce the shuffling load. To our best knowledge, existing works on the CDC only consider input files with uniform file size, limiting their practicality in real-world applications. To address this limitation, we propose a Heterogeneous Coded Distributed Computing (HetCDC) scheme to handle input files of nonuniform sizes. We then formulate a joint optimization problem to optimize the file placement and coded shuffling strategies to minimize the shuffling load. Through reformulation, we convert the nonconvex optimization problem into an integer linear programming problem and solve it through the branch-and-cut method. Numerical studies show the proposed HetCDC outperforms existing works. Based on the Het- CDC, we further develop a Heterogeneous TeraSort algorithm to improve the sorting time of traditional TeraSort, which is a key building blocks for many big data processing algorithms.
ADHD and career sustainability: A sustainable career ecosystem perspective
(Emerald, 2025-09-26) Grabarski, Mirit K.; Jameson, Tiffany Payton; Mouratidou, Maria
Purpose: We explore the perceptions of career sustainability of individuals with attention-deficit/hyperactivity disorder (ADHD) in the United States, taking a sustainable career ecosystem perspective that considers multiple sustainability indicators and different interdependent actors. Design/methodology/approach: We conducted semi-structured interviews with 31 participants and analyzed the data using a template approach that allows combining deductive and inductive analysis. Findings: We identify how ADHD impacts different aspects of sustainable careers, namely time, person-related factors and indicators (i.e. happiness, productivity and health). Moreover, our findings identify empirical support for two additional indicators (financial security and growth mindset) as proposed by sustainable career ecosystem theory. We suggest a disproportionate impact of ADHD on the indicators, specifically, productivity, due to contextual workplace barriers. We also identify key actors at the local ecosystem level (e.g. family members, teachers, neighbors, friends, co-workers and therapists) that play an important role in individual careers within the ecosystem, particularly regarding diagnosis and support. Originality/value: We provide empirical insights that support the recently developed sustainable career ecosystem theory and suggest a differential impact of ADHD on the indicators.
Technology-facilitated sexual violence: exploring vulnerability factors, distress, and coping strategies in women
(2025) Oliver, Casey L.; Mazmanian, Dwight; Musquash, Aislin; McQueen, Karen; Barata, Paula
Technology-facilitated sexual violence (TFSV) — sexual abuse perpetrated through technological means — consists of several forms of unwanted sexual behaviours, including sexual harassment, cyberstalking, image-based exploitation, unwanted sexual experiences, and gender and sexuality-based harassment. Although research on this form of gender-based violence has expanded in recent years, little is known about the women who experience it. This dissertation aimed to examine multiple dimensions of victimization to better understand TFSV. Study One used survey data to identify demographic and technological differences that may constitute vulnerabilities in women victims (N = 268). Young age, sexual minority status, popularity online, and certain patterns in technology use were identified as potential vulnerabilities for women. Study Two investigated the temporal relationships between TFSV and distress in women (N = 55) using a smartphone-based ecological momentary assessment paradigm. The results revealed multidirectional trends between these variables, including that TFSV may influence victims’ negative and positive affect, distress may change over time due to TFSV experiences, and distress may also serve as a vulnerability factor for TFSV. Study Three explored how women cope with TFSV in its aftermath. Through in-depth interviews, women (N = 9) described using cognitive/emotional and behavioural risk avoidance strategies to protect themselves, including staying on guard, altering their appearance, disappearing into the shadows, assessing strangers, and protecting other women. Overall, this project illuminated the experiences of women victims of TFSV, with the goal of informing prevention and intervention strategies that center their experiences alongside the broader literature.
Simultaneous broadband vibration suppression and energy harvesting using a magnetically enhanced piecewise-linear nonlinear energy sink
(2025) Li, Haining; Liu, Kefu; Deng, Jian; Elshaer, Ahmed; Khalid, Muhammad
Nonlinear energy sinks (NESs) offer significant potential for simultaneous broadband vibration suppression (VS) and energy harvesting (EH) through the use of an essentially nonlinear spring (ENS). However, realizing an ENS with minimal energy dissipation remains challenging. A piecewise-linear spring (PLS) provides a structurally simple and physically interpretable means to approximate nonlinear stiffness with little added friction. Yet, NESs employing a PLS often require a relatively high initial energy threshold to trigger targeted energy transfer (TET), resulting in reduced performance under low-level excitation. Magnetic springs can introduce bi-stable characteristics that enable snap-through oscillations, thereby lowering the energy threshold. Existing studies, however, have focused primarily on ungrounded magnetic spring configurations, leaving the influence of a grounded magnetic spring (GMS) on NES’s performance largely unexplored. To address this gap, this research integrates a tunable GMS into a piecewise-linear NES (PLNES) to reduce the energy threshold, thereby facilitating TET activation and enhancing broadband VS and EH performance. Four interconnected studies are undertaken: 1. Magnetic Spring Modelling – A tunable multi-stable piezoelectric energy harvester (PEH) is developed by combining a cantilever beam with an adjustable magnetic assembly capable of achieving mono-, bi-, and tri-stable states. Two magnetic restoring force models, based on the magnetic single-point and two-point dipole approaches, are formulated and optimized via a multi-population genetic algorithm. Parametric sensitivity analyses are conducted for the optimal models. 2. Hybrid Multi-Stable Energy Harvesting – A multi-stable hybrid energy harvester (MSHEH), integrating a PEH and electromagnetic energy harvester (EMEH), is proposed and evaluated numerically and experimentally under various stability states. Optimal load resistances for balanced energy output across configurations are determined through optimization. 3. PLS Design Methodology – A systematic approach for designing a PLS is developed, enabling close emulation of a desired ENS using a cantilever beam constrained by single- or double-stop blocks. The designed PLSs are validated against the target ENS through both simulation and experiment. 4. Magnetically Enhanced PLNES (MPLNES) – A novel MPLNES is proposed by integrating a PLNES with a tunable GMS and a grounded EMEH. The GMS produces a position-dependent restoring force that shifts the NES’s equilibrium, enabling easier activation of large-amplitude oscillations. Numerical and experimental results confirm that the MPLNES triggers TET at lower excitation levels than the corresponding PLNES. A two-objective optimization reveals that the MPLNES achieves superior trade-offs between VS and EH, sustaining energy transfer over a wider range of excitation levels compared with the two other NES designs. Keywords Nonlinear energy sink; vibration suppression; energy harvesting; targeted energy transfer; essentially nonlinear spring; piecewise-linear spring; grounded magnetic spring; piezoelectric energy harvester; electromagnetic energy harvester; multi-stable dynamics.