Skip to main content
ARS Home » Plains Area » Temple, Texas » Grassland Soil and Water Research Laboratory » Research » Publications at this Location » Publication #400139
Research Project: Development of Enhanced Tools and Management Strategies to Support Sustainable Agricultural Systems and Water Quality

Location: Grassland Soil and Water Research Laboratory

Title: Using UAS-multispectral images to predict cord yield under different planting dates

Author
item SAPKOTA, BALA - Texas A&M University
item SARKAR, SAYANTAN - Texas A&M Agrilife
item BAATH, GURJINDER - Texas A&M Agrilife
item Flynn, Kyle
item Smith, Douglas

Submitted to: ASA-CSSA-SSSA Annual Meeting Abstracts
Publication Type: Abstract Only
Publication Acceptance Date: 6/20/2022
Publication Date: 11/7/2022
Citation: Sapkota, B., Sarkar, S., Baath, G.S., Flynn, K.C., Smith, D.R. 2022. Using UAS-multispectral images to predict cord yield under different planting dates. ASA-CSSA-SSSA Annual Meeting Abstracts [abstract], Baltimore, MD, November 6-9, 2022.

Interpretive Summary: Seedling emergence is a crucial agronomic factor for making field decisions such as replanting, which is time sensitive. Conventional methods to determine emergence using manual stand count is time and resource consuming. Recent aerial imagery based systems using unmanned aerial vehicles (UAVs) is a high-throughput way to scout fields. However, the system is based on discriminating green pixels of plants from soil. This method is unreliable for organic production systems which might consist of weeds and crop stubbles along with seedlings. Therefore, the goal of our study was to develop an algorithm to estimate cotton seedling count from aerial imagery after automatically discriminating them against weeds and stubbles. The methods here utilized automated plant detection and counting algorithms that were programmed in Python. Seedlings counted aerially had accuracy of 70%. Our method can help in high-throughput estimation of seedlings in organic production systems where no herbicide application and zero till plantings are practiced.

Technical Abstract: Seedling emergence is a crucial agronomic factor for making field decisions such as replanting, which is time sensitive. Conventional methods to determine emergence using manual stand count is time and resource consuming. Recent aerial imagery based systems using unmanned aerial vehicles (UAVs) is a high-throughput way to scout fields. However, the system is based on discriminating green pixels of plants from soil. This method is unreliable for organic production systems which might consist of weeds and crop stubbles along with seedlings. Therefore, the goal of our study was to develop an algorithm to estimate cotton seedling count from aerial imagery after automatically discriminating them against weeds and stubbles. Aerial imagery was collected using MicaSense RedEdgeMX dual camera system mounted on a DJI Matrice 600 UAV. Images were orthomosaiced using Pix4D mapper and required rasters were created in ArcGIS Pro. Automated plant detection and counting algorithms were programmed in Python. Seedlings counted aerially had accuracy of 70% (R2 = 0.7). Our method can help in high-throughput estimation of seedlings in organic production systems where no herbicide application and zero till plantings are practiced.