Purdue University Graduate School
Browse

META AG: AN AUTOMATIC CONTEXTUAL AGRICULTURAL METADATA COLLECTION APP

Download (2.91 MB)
thesis
posted on 2023-04-29, 02:27 authored by Md Samiul BasirMd Samiul Basir

  

Data is the foundation of digital agriculture. Data from a wide variety of sensors in the soil, in machinery, or from remote sensing can inform decisions including site-specific land and crop management but capitalizing on these data requires metadata that captures the full story related to production. Answers to metadata questions such as who, what, where, when, and how are often unavailable when aggregated data are analyzed. These metadata are crucial for making accurate operation and management decisions and certainly for developing AI models. Since farmers and researchers exhibit human behavior of forgetting to take notes or entering incorrect information, even with digital means, missing and erroneous records are common. To address this issue, a metadata collection Android app – Meta Ag for agricultural activities was created that automatically appends the operator’s name, time, and space information to an in-field event, and provides a user-friendly interface to gather information with more details describing which activity was done and how. Meta Ag has six main modules, including user registration, geofence construction, accessed geofence recognition, an infobot for extensive activity data collection, setting options for Infobot and data access. By design, manual data input, with automatic validation, when possible, was used for information collection. To achieve this, Meta Ag uses dynamically constructed, interactive option lists for fewer data entry errors. The collected data were stored in a Google Firebase database as central storage for multiple users. To facilitate data interoperability, stored data were made accessible in CSV and JSON format. The Android app collects extensive metadata from database interactive option lists and the infobot as a data collection wizard provides a dynamic environment for data collection in a short time with minimal manual input. The app was also able to reduce missing data as it automatically records the accessed fields and activity time in that field. The Meta-Ag app can work both as a standalone tool and an integration into any other farm information management system.

Funding

NSF grant EEC-1941529

History

Degree Type

  • Master of Science in Agricultural and Biological Engineering

Department

  • Agricultural and Biological Engineering

Campus location

  • West Lafayette

Advisor/Supervisor/Committee Chair

Dennis R. Buckmaster

Additional Committee Member 2

James V. Krogmeier

Additional Committee Member 3

Ankita Raturi

Additional Committee Member 4

Bruce J. Erickson

Usage metrics

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC