22505 research outputs found
Sort by
Predicting Hollywood Movie success using Predictive Machine Learning Algorithms
No full textThe Decision Trees are always simple to understand and interpret techniques, a single tree may not be enough for the model to learn the characteristics from. In this IMDb movie review prediction problem, evading all other simple mechanisms, the Random Forest, on the other hand, is a Tree -based algorithm that makes judgments by combining the attributes of numerous Decision Trees is used. The primary goal of this work is to evaluate the predictive performance of a random forest model with various parameters used for forecasting numerical user ratings of a movie based on pre-release data such as actors, directors, profit, social media reviews, and movie genres. Although a slight difference has been indicated by the results of the two variant models, one should also note that both these models show great similarities in terms of their prediction performance, making it hard to draw any general conclusions on which model yields the most accurate movie predictions
Intelligent Monitoring and Alerting System using Some Deep Learning Models
No full textRoad traffic crashes cause over 1.19 million fatalities per year worldwide and represent a major challenge for urban safety (World Health Organization, 2023; World Bank Group & WRI India, 2020). Smart city initiatives aim to use technology to reduce such losses through rapid accident detection and response (Arefin et al., 2025). This proposal outlines a vision for a Road Accident Detection and Response System that employs computer vision and deep learning to identify accidents from video feeds and immediately notify emergency services. The core of the system will be a custom-trained YOLOv8 object detection model, leveraging a dataset of annotated accident images (cars, buses, trucks, motorcycles) from Roboflow Universe (Geetha et al., 2024). Exploratory Data Analysis (EDA) will characterize the dataset (class distribution, bounding box sizes, environmental conditions) to guide training. The model will be developed locally on an NVIDIA RTX 2060 GPU using the Ultralytics YOLOv8 framework (Ultralytics, 2023). A data.yaml file will define the custom accident classes and training/validation splits, following YOLO conventions (Ultralytics, 2023). After training, the system’s performance will be rigorously evaluated on held-out data by computing precision, recall, F1-score, and mean Average Precision (mAP) at various IoU thresholds (Geetha et al., 2024). These metrics will quantify detection accuracy (for instance, YOLOv3 achieved approximately 57.9% [email protected] in 22 milliseconds of inference time) (Redmon and Farhadi, 2018), confirming real-time viability. The final system will integrate the detection model with an alert mechanism to dispatch notifications (e.g., SMS, email) to first responders or traffic management centers (Arefin et al., 2025). Key deliverables include the trained model, a codebase for real-time accident detection, a detailed analysis of dataset and model performance, and a documented prototype implementation. This work fills a gap in smart-city safety by applying the latest YOLOv8 technology to automated accident detection and rapid notification (Khalili & Smyth, 2024). It builds on recent research showing YOLOv8’s superior accuracy and speed (Ultralytics, 2023; Ghazzaoui and Kubra, 2025), but extends it to the critical domain of traffic incident response, aiming to reduce emergency response times and save lives. Recent work also reports strong accident-detection results with an improved YOLO11 (‘YOLO11-AMF’) that adds a linear-attention module, an asymptotic FPN, and a Focaler-IoU loss, improving precision/recall and mAP over the YOLO11n baseline on a curated accident dataset (Li, Huang & Lai, 2025
The World is Beautiful
No full textMy thesis, The World is Beautiful is a short, digitally animated film that follows the journey of a jaded employee working at a dystopian fishery. The intention of this piece was to tell a well-paced, coherent story that also served as an allegory to spark verbal discourse in an audience. As a surface level story, the film portrays a working protagonist who has been forced to remain seated at her desk by a mechanical chair. She, presumably, has been monotonously pressing icons on a screen for her entire life until something catches her eye and prompts her to break free from her desk and seek escape. Allegorically, the film was made to promote conversation about the role of the human creative spirit existing in a capitalist environment. The specifics of my plot and design choices would undergo many changes throughout the filmmaking process as my film’s metaphorical significance continued to be honed. Treating the film as an allegory would prove to serve as the engine and stopper for much of my creative process, since the challenge of assigning meaning to each small decision (while still trying to tell a straight forward story) meant a lot of additional responsibility to myself as an artist. In the end, the combination of many drafted treatments and storyboards, deep conversations with committee members and peers, and a cementing of confidence in myself as a creator served to make a piece that I believe ultimately accomplished my goal. Clarivate log
Accelerating the degradation of biodegradable mulch films in soil and compost environments
No full textThe incomplete degradation of biodegradable mulch films (BMFs) in agricultural soils poses environmental challenges, with persistent plastic residues impacting soil health. This dissertation investigates the use of bioaugmentation with P. guariconensis to enhance BMF degradation under laboratory, raised bed, and field conditions. The research focuses on optimizing microbial delivery methods, evaluating the efficacy of drip and spray bioaugmentation techniques, and assessing mass loss as a primary metric for degradation. Laboratory experiments established proof-of-concept by demonstrating enhanced carbon mineralization, weight loss, and fragmentation of BMFs under bioaugmented conditions. These findings were translated to field-scale trials, where bioaugmentation treatments consistently outperformed untreated controls in terms of mass loss. Among all treatments, the highest bioaugmentation treatment, which combined inoculation 15 days before tillage and 7 days before tillage using both drip and spray methods, achieved the most significant degradation, with near-complete breakdown of the BMF observed in some instances. Raised bed trials provided a controlled environment to validate field results, showing consistent trends in accelerated degradation for both Multi Layered Agricultural Mulch Film (MLAMF) and EcoVio-2 (EV2) films. Despite these outcomes, a one-way ANOVA revealed no statistically significant differences between treated and untreated sample masses after one year, likely due to variability introduced by field conditions such as soil heterogeneity, environmental fluctuations, and microbial competition. Mass loss was the primary metric for evaluating biodegradation, offering practical insights into film breakdown. However, the study highlights the limitations of this approach and emphasizes the need for complementary techniques, such as enzymatic assays and carbon mineralization analyses, to directly measure microbial activity and quantify biodegradation. For example, monitoring plastic-degrading enzymes like depolymerases could validate the mechanistic role of bioaugmentation, while carbon mineralization assays could better link microbial processes to observed mass loss. This work demonstrates the promise of bioaugmentation as a scalable and sustainable strategy for managing BMF residues in agricultural soils. While challenges remain in achieving consistent results under real-world conditions, this research lays the groundwork for integrating bioaugmentation into agricultural practices to mitigate the environmental impact of biodegradable plastics
Adaptive Reuse of Underutilized Strip Malls: Creating Mixed-Use Continuum Care Campuses for Seniors
No full textAs suburban commercial landscapes continue to decline, underutilized strip malls present an opportunity to reimagine aging-supportive environments through adaptive reuse. This thesis investigates how these sites can be transformed into mixed-use senior housing and continuum-care campuses through adaptive reuse strategies. Using a comparative case study methodology, the research examines a range of precedents—including senior housing retrofits, dementia villages, and community-based mixed-use developments—to identify key spatial, programmatic, and environmental principles that support aging populations. A five-part evaluative framework assesses project performance across site integration and green space, social and intergenerational engagement, programmatic and service integration, mixed-use activation, and cognitive and environmental support. These criteria apply to selected case studies and are further substantiated through interviews with architects, professors, and senior living professionals. Findings indicate that successful adaptive reuse relies on flexible spatial organization, strong community connectivity, and evidence-based design integration. The proposed framework outlines strategies that respond to the growing demand for senior housing while advancing broader goals of suburban revitalization. The study concludes by defining design approaches that integrate architectural, planning, and age-inclusive principles to guide future suburban strip mall retrofits and senior care development
Understanding the Interplay Between the Digital and the Physical in Shared Augmented Reality Gaming: Probing through Urban Legends
No full textShared Augmented Reality (Shared AR) is an emerging technology that enables multiple users to interact synchronously within a collocated AR environment. Yet, there is limited research on the group interactions and dynamics in Shared AR, particularly in the context of gaming. To address this gap, we investigate Shared AR group interactions using a phone-based Shared AR mobile game called Urban Legends. Through in-situ observations, focus groups, and one-on-one interviews with 22 participants, we examine how users collaborate and communicate within the game. Our findings reveal that while verbal communication predominates, non-verbal cues are often overlooked by collocated participants, and users initially struggle to recognize the expansive virtual space and the need for physical movement. Over time, users adapt to the hybrid environment, demonstrating increasing spatial awareness and more dynamic collaboration. Based on these insights, we present a suite of design recommendations for enhancing spatial awareness, supporting multi-modal communication, and fostering engaging group dynamics in future Shared AR applications
Rooted in Ogdensburg: A Creative Exhibition from Classroom to Gallery
No full textPicture this: you are an elementary student- wandering around an art exhibition. On the wall hangs artwork you dedicated your time to. How does it make you feel? Art exhibitions of student work are a display practice implemented at the James F. Montpelier Gallery located at Ogdensburg Free Academy in Ogdensburg, New York. Such student-focused art exhibitions, historically, create a wonderful, inspirational, and impactful environment for students to display their finest works. Once the COVID-19 pandemic hit the world by storm in 2020, the thematic shows shifted to assemblages of students’ best work. This is due to the limited time teachers had to put a show together, social distancing, lack of supplies, and the preparation of another widespread quarantine, like the one in 2020. However, this exhibition aims to revitalize the curated, themed, public art exhibition to emphasize the ability of an exhibition to boost student participation, self-esteem, and passion within the arts. The exhibition, “Finding Balance: Exploring our Relationship with Nature” will serve as a case study to answer the question: “How can a pre-K-12 art exhibition at Ogdensburg Free Academy influence students’ self-esteem and sense of identity through creative processes and peer recognition?” Researching my topic requires three main methodologies, including primary and secondary research on local demographics; history, and culture; the planning and execution of an exhibit; and community-engaged artistic practice. Involving student and teacher interviews, workshops, and projects. This research and project will culminate in the opening of the exhibition “Finding Balance: Exploring our Relationship with Nature” on May 1st, 2025, at the James F. Montpelier Gallery located at Ogdensburg Free Academy. Curating this show engages and celebrates students\u27 art, gives the community a physical example of their learning progress, and validates students\u27 hard work and dedication
The Impact of the 2008 Financial Crisis on Crime and Cybercrime in the United States
No full textThis thesis will examine how the traditional crime and cy- were affected by the 2008 financial crisis. United States are also experiencing a rise in crime between 2005 and 2012. Based on the FBI data at the national level. The Internet Crime Complaint Center (IC3), Uniform Crime Reports and important economic indicators. The study, which involves tors, including unemployment, GDP, rates of foreclosures, and mortgage rates, is a combination. correlationbased interpretation supported by exploratory descriptive trend analysis. model-fit checks. The results indicate that contrary to the conventional expectations, violent and property crime also maintained their long-term reduction during the period of the crisis, but cybercrime. grievances and other financial losses increased tremendously. Crime Traditional crime displays pre- unanimously negative relations with unemployment and foreclosure measures, consistent. and opportunity based and security based explanations, but cybercrime is positively correlated. as a country with economic hardship and increased reliance on the digital. Considering that the number of annuals is low. observations in the national series, procedures of predictive claims in the future do not come into consideration of this study, and high in-samples fit should be taken with reservations. The findings are all in favour of. strain-opportunity and routine activity approaches and accentuate the necessity of crisis-era. policies that have a combination of cybersecurity, digital literacy, and economic assistance to minimize both. offline and online harms. Main keywords: financial crisis of 2008, traditional crime; cybercrime, economic indicators, Machine learning, time-series analysis, United States, Routine activity theory
StrainX : A Technical Exploration of Cinema4D\u27s Native Particle System in an Experimental Title Sequence
No full textThis paper presents a practice-based investigation into the capabilities of Cinema 4D’s native particle system through the creation of an experimental title sequence titled StrainX. The project was conceived as a technical and aesthetic exploration, prioritizing abstraction, form, and motion over narrative content. Through iterative testing and simulation, the study examines how Cinema 4D’s built-in particle, geometry, and material tools can generate visually complex results without relying on third-party plugins, such as X-Particles. All assets, simulations, and textures were developed natively within Cinema 4D, allowing for a unified and efficient workflow. The methodology emphasizes procedural experimentation, simulation caching using Alembic files, and optimization for iterative rendering. The findings suggest that the native particle system is both accessible and capable of supporting sophisticated motion design outcomes, enabling faster iteration, reduced workflow fragmentation, and greater stability. The paper also reflects on the project’s educational relevance and its implications for future motion design practices
Macroscale Thermal Modelling for the Molten Metal Droplet Jetting Additive Manufacturing Process
No full textMolten Metal Jetting (MMJ) has become a promising pathway for next-generation metal additive manufacturing because it avoids the need for powders or high-power energy sources and enables precise, drop-on-demand deposition of structural alloys. Despite this promise, the field lacks a predictive thermal modeling framework that can describe the full temperature history of a part as thousands of droplets accumulate. Without such a model, it is difficult to control heat buildup, to understand how bonding conditions evolve across layers, and to avoid defects such as incomplete fusion, porosity, and geometrical distortion. This dissertation presents the first validated macroscale thermal modeling framework for the MMJ process, combining experimental characterization, analytical modeling, and a graph-theory numerical approach to capture the full spatiotemporal thermal behavior of MMJ builds. The work begins with a detailed review of single-droplet physics, thermal interactions, and heat accumulation phenomena, followed by an experimental study performed on the Xerox ElemX system that measures top-surface temperatures for one, three, and five simultaneous builds. These measurements reveal how interlayer cooling time increases when multiple parts are printed in parallel, and they provide a controlled dataset for validating the proposed model. Additional thermocouple measurements on printed aluminum parts quantify the influence of natural convection, platform motion, argon shielding flow, and printhead proximity. These data are used to compute realistic convective heat transfer coefficients, addressing a long-standing gap in MMJ modeling where convective losses were previously simplified or ignored. Using this experimental foundation, the dissertation develops an analytical model based on fin theory to identify dominant cooling mechanisms and to determine height-dependent convection coefficients for MMJ. Building on these results, a graph-theory thermal model is created to simulate part-scale heat conduction during sequential droplet deposition. The model represents the part as a network of thermal nodes that exchange heat through conductive pathways while losing heat through spatially varying convection. Sequential droplet inputs are superimposed in time to generate a fully resolved thermal history without the computational burden of high-fidelity CFD or conventional finite element simulations. Model predictions are validated against thermal camera measurements for single-part, three-part, and five-part cases. The comparisons show that single-part builds experience a gradual surface temperature drop as a function of part height due to minimal cooling time between layers, while multi-part build surface temperatures cool down more quickly due to extended interlayer delays. The model captures these trends accurately and predicts cooling rates, temperature gradients, and peak temperatures across full part heights. The results also highlight how droplet frequency, build layout, and convection strength jointly control the thermal state that governs droplet spreading, remelting behavior, and microstructural development. This dissertation provides three primary contributions to the field: the first part-scale thermal model for MMJ, the first experimentally grounded convection analysis for MMJ environments, and the first demonstration that build sequencing can be used intentionally to modulate interlayer cooling. The framework presented here establishes a foundation for predictive process planning, supports future microstructure modeling, and enables the development of real-time control strategies for molten metal jetting