Utilizing QGIS for open-source UAS imagery plant classification and plot segmentation

The analysis of unoccupied aerial system (UAS) imagery remains a bottleneck to obtaining actionable data due to its complexity and required specialized skill set. We introduce a workflow designed to balance usability and flexibility in processing UAS imagery using QGIS, an open-source geographic information system software. The workflow consists of four sequential steps: semi-automated plant classification, user-corrected plot delineation, spectral index calculation, and data extraction. We tested this workflow on barley UAS image data, collected over 12 flights throughout the growing season. As a proof-of-concept, multiple red-green-blue (RGB) indices were calculated to test for relationships to ground-sampled data and exemplify applications of spectral analysis. Our results indicate a strong correlation between the visible atmospheric resistant index and ground truth data, particularly when integrated over multiple flights. This open-source workflow provides a lower barrier-to-entry solution for researchers and producers, facilitating the broader adoption of UAS technology in agriculture. By automating routine tasks while allowing user intervention for critical adjustments, this approach enhances the efficiency of agricultural data analysis.

Back to Home page