The goal of this project is to build technology and establish transfer pathways to enable commercialization of low-cost precision agriculture technologies that automate monitoring, analysis, and mapping of soybeans, cover crops, and weeds in US soybean production for farmers and researchers. In the first year (FY22) of this five-year project we 1) built cyberinfrastructure needed for a weed image repository; 2) constructed cloud computing pipelines for automated 3-dimensional analysis of plant images for density and biomass estimation; and 3) tested and calibrated a low-cost sensing platform (GoPro camera) for real-time mapping of weed species density and biomass. In the second year of this project (FY23) we deployed a new camera system (OAK-D) that provides better resolution of depth maps used for plant biomass estimation and expanded the image repository and calibration of the sensing technology to include soybeans and cover crops. These activities grew the body of knowledge to inform new practices, tools, and uses through continued research and development of computer vision technology for precision soybean production. Success in FY23 prepared us to rapidly scale-up the technology for a variety of research and production applications. By the end of the five year project we will translate our findings into refined tools that can be released to farmers and private industry, supporting the sustainability of U.S. soybean production. Stakeholder engagement and accelerated adoption of the technology continued to be supported by the Getting Rid of Weeds (GROW) team’s outreach and extension specialists.

Our goal is to build technology and establish transfer pathways to enable commercialization of low-cost precision agriculture technologies that automate monitoring, analysis, and mapping of soybeans, cover crops, and weeds in US soybean production for farmers and researchers. In years one and two (FY22, FY23) we 1) built cyberinfrastructure needed for a weed image repository; 2) constructed cloud computing pipelines for automated 3-dimensional analysis of plant images for density and biomass estimation; 3) tested and calibrated low-cost sensing platforms for real-time mapping of weed species density and biomass; and 4) expanded the image repository and calibration of the sensing technology to include soybeans and cover crops. In FY24 we will 1) continue to develop camera variants for multiple platforms and operationalize computer vision systems for research and production; 2) develop user-friendly user interfaces for researchers (to be followed by same for farmers); 3) continue calibration of cameras through the imaging of soybean cultivars and cover crop individual species and mixtures and via the manual counting, identification, and weighing of weeds and soybeans in diverse environments to create training data sets for artificial intelligence; 4) expand on previous camera calibration work with a focus on cover crop biomass, species composition, and effect on nutrient availability; and 5) continue outreach work. These activities will grow the body of knowledge and inform new practices, tools, and uses through continued research and development of computer vision technology for precision soybean production. Success in FY24 will prepare us to rapidly scale-up the technology for a variety of research and production applications. By the end of the five year project we will translate our findings into refined tools that can be released to farmers and private industry, supporting the sustainability of US soybean production. Stakeholder engagement and accelerated adoption of the technology will be supported by the GROW team’s outreach and extension specialists.