Creative Components
Permanent URI for this collection
Browse
Search Collection
1 - 5 of 1764
-
TextThe Future of The Superbowl( 2024-08)This research report aims to find ways in which the Superbowl can be made more sustainable and to use those findings to implement them not only into the Superbowl but other major events in years to come. The Superbowl is one of the largest events in America and is hosted at the same time every year in different cities across the US. This event brings people from all over to a select location for about a week in advance of the game and normally a day or two after the event has ended. I would like to locate areas in which improvement is necessary as well as areas in which improvement may not be necessary but is still available.
-
Learning ObjectVisium's CytAssist, An Optimized Protocol for the Bell Lab( 2024-08)This work should serve as a reference guide and learning tool for the CytAssist technology by 10xGenomic's, which has been optimized for use at Iowa State College of Veterinary Medicine's Department of Veterinary Anatomic Pathology.
-
Technical ReportIntelligent Table Transcriber: AI-driven Table Recognition Framework for Enhanced Information Retrieval( 2024-08)Data information retrieval aims to extract useful information with the query from the users. Traditional information retrieval systems often struggle with diverse document formats, hindering e cient data extraction and organization. Motivated by these limitations, this project collaborates with Soilserdem, aiming to design a framework for data information retrieval from a variety of data types with agronomic information to a machine-readable format. (e.g., PDF, images, geospatial files). In particular, the main function of this project includes recognizing tables and extracting data from digital images. One challenge was to capture the textual content while preserving the original tabular layout when directly applying the existing table structure recognition model. In this creative component, we propose (1) a hybrid approach that feasibly captures text content and preserves the original tabular layout by combining a pre-trained table transformer and a rule-based approach with optical character recognition. (2) Additionally, we design the overall back-end framework, from data loading to main processes. Our exploratory experiments demonstrate that the proposed hybrid approach better captures textual content and preserves the original tabular layout than a single model of rule-based and data-driven approaches.
-
PresentationBeginners Guide to Regenerative Agriculture and Why It Matters( 2024-08)Modern agriculture has been built on the advancement of technology. Though these industrialized practices have increased the yield of many crops, concerns with the sustainability of certain practices have been raised. Some of these concerns are the environmental impact, resource depletion, impact on biodiversity, food quality, long term food security, and societal issues. Regenerative agriculture offers an alternative approach to standard conventional systems. Regenerative agriculture is an approach that focuses on soil health and biodiversity to work with natural systems. These systems show promise with carbon sequestration, water conservation, and climate resilience. Though a more intensive system, the profitability of these systems can be a net positive for growers, as well as boost local economies through job creation. Regenerative agriculture is a farming paradigm based on soil health principles. This approach to farming aims to build the soil and support or simulate natural ecosystems to sustain crop growth. The principles of soil health foster the essential functions of the soil by supporting both biotic and abiotic systems. This creative component will act as a guide for both growers considering the adoption of regenerative agricultural practices and consumers interested in learning more about regenerative agriculture and for supporting its management approach. This guide intentionally provides a broad overview, serving as a starting point for interested parties, explaining the basics about the soil, and how regenerative agriculture supports healthy ecosystem functions. This work does not exist to validate or disprove concerns with current agricultural systems, rather it empowers readers with information to make informed farming decisions. Anyone interested in using this information is advised to conduct their own local research to better understand regenerative practices best suited for their location and cropping systems. The results of regenerative agricultural practices may vary depending on cropping systems and specific geographic locations. Literature review was used to compile field research data, growers’ testimonies, and the ethical and intellectual insight posed by parties interested in regenerative practices for this guide. Personal photos and experiences will also be shared throughout this work, as I share the story of my own farming experiences that began in the fall of 2021. This creative component primarily focuses on soil science, sustainability, and human health benefits, while aiming to present thought-provoking ideas on alternative systems of agriculture. It was written for consumers, producers, and other interested parties. The goal of this is to begin with the basic knowledge of different areas of soil science and expand into regenerative agricultural practices and the scientific research that supports beneficial claims.
-
Technical ReportFrom Static Scans to Dynamic Text: A Transcription Transformation( 2024-08)This creative component introduces an intelligent transcriber that automates the categorization, extraction, and digitization of text, tables, and form data from scanned copies, images, and geospatial files. The system employs a Naive Bayes classifier Vikramkumar et al. (2014) to classify the files into predefined categories based on the file name or metadata in the case of geospatial files. The classifier’s prediction is confirmed with the user, and if correct, the file is assigned to the respective category. In case of an incorrect prediction, the user provides the correct category, and the Naive Bayes classifier is updated accordingly. We investigated the idea of using OCR tools to help classify, digitize, and extract useful information from a large amount of paper documents. This functionality is meant to be integrated into an online dashboard that AgTech startup Soilserdem is building for its customer: a dashboard that can function as a hub for all farming-related documentation. We surveyed the literature for free, open-source, yet state-of-the-art tools such as Tesseract and Keras-OCR to help build this platform. We implemented both infrastructural requirements (API endpoints, docker containers, database changes) in addition to the core transcription logic. Due to time constraints and the complexity of the problem, we focused only on a subset of document types. The transcriber utilizes advanced deep learning and computer vision techniques to accurately recognize and convert the extracted information into appropriate digital formats, such as Excel sheets for tabular data and key-value pairs for form data. For complex scanned copies, the transcriber generates an HTML file that faithfully reproduces the original document’s layout using text boxes and precise text coordinates. A combination of template matching and transcription using Keras-OCR was used in our final implementation to ensure standardized output and maximize accuracy. We achieved a best-case scenario accuracy of 75% for the transcription of handwritten text. Additionally, the system incorporates a search feature that enables efficient retrieval of relevant documents by extracting and indexing the text content from the processed files. This innovative solution streamlines the conversion of unstructured data into searchable and easily manageable digital assets, enhancing productivity and information accessibility across various domains.