Description

VegClassify is a hybrid, high-resolution greenspace classification tool integrating a fine-tuned deep learning model with automated NDVI threshold optimization. The tool improves accuracy and reduces subjectivity in vegetation classification using NAIP imagery. This graphical user interface (GUI) was developed to extend and simplify the use of the previously established greenspace image classification tool. The GUI provides an accessible, user‑friendly environment for running the VegClassify workflow without requiring users to interact directly with Python scripts or command‑line tools. It supports loading NAIP or other high‑resolution imagery, running deep‑learning–assisted NDVI threshold optimization, visualizing classification outputs, and exporting vegetation masks.

The underlying Python codebase that implements the VegClassify hybrid deep learning–NDVI threshold method was originally published in:

Wang, H., Zhao, X., Gholami, S., McGinty, C., Chamberlain, B., & Qi, X. (2026). A Hybrid Deep Learning and NDVI Threshold Approach for High-Resolution Urban Greenspace Classification. Urban Forestry & Urban Greening, 129332.

This GUI serves as an application‑layer wrapper for that Python tool, enabling broader accessibility for research, planning, education, and applied landscape analysis.

Author ORCID Identifier

Huaqing Wang https://orcid.org/0000-0002-1630-9753

Document Type

Dataset

DCMI Type

Dataset

File Format

.exe

Viewing Instructions

VegClassify.exe - This is the standalone GUI application for the VegClassify greenspace classification tool. The executable incorporates all required Python packages and dependencies, so no installation is necessary. Users can simply double‑click the file to launch the program, and the GUI will start running immediately. It provides an accessible interface for loading imagery, executing the VegClassify workflow, visualizing classification outputs, and exporting vegetation masks.

Publication Date

3-12-2026

Publisher

Utah State University

Methodology

NAIP 0.6-m R/G/B/NIR imagery was sampled across 48 U.S. states. Deep learning model was fine-tuned, and NDVI thresholds were evaluated to determine optimal segmentation.

Language

eng

Code Lists

see README

Disciplines

Environmental Design | Landscape Architecture

License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Checksum

35ef9df86aba641fc4b98d7c56b90b83

Download

Additional Files

README_VegClassify.txt (2 kB)

Share

 
COinS