Advanced search
Start date

Equipment for localized herbicides spraying for weed control in pre-planting and after soybean crop emergence using multispectral images and computer vision


Currently, Precision Agriculture (PA) is an area that has been receiving continuous investments in the country due to its potential to optimize agricultural inputs and productivity through data analysis, collected by sensors or implements. PA assists farmers in decision-making processes, enabling site-specific crop management considering the heterogeneity of plots and spatial-temporal variabilities. Total area chemical control of weeds using herbicides leads to significant environmental pollution, high costs for farmers purchasing these inputs, and unintended exposure of non-target organisms, contributing to herbicide resistance in weeds widely used in the field. To address these issues, localized spraying equipment designed to be installed in agricultural machinery has emerged as a promising alternative. These devices can apply herbicides solely to detected weeds, generating interest among the scientific-business community and numerous farmers worldwide. This research project aims to develop a prototype of such equipment capable of real-time detection and classification of weeds and soybean plants using Artificial Intelligence and multispectral images for installation on agricultural sprayers. The goal is to enable site-specific herbicide applications during different stages of crop management, such as desiccation and post-emergence. Unlike other solutions in the market, the proposed system stands out for its utilization of multispectral images acquired through the combination of two cameras: one color camera (RGB) and one monochromatic camera with a pass-band filter in the near-infrared (NIR) wavelength. Additionally, the system employs state-of-the-art Artificial Intelligence algorithms for detection and classification tasks, driving forward Agriculture 4.0 in the country. Furthermore, the system will be the first equipment commercially available in Brazil to effectively perform herbicide applications after soybean crop emergence in the field. This capability is achieved by leveraging a multispectral image database built in a soybean plantation during Phase I, which allows the developed AI algorithms to distinguish soybean plants from weeds. The system will also differentiate between narrow-leaf and broad-leaf weeds, enabling farmers to use more suitable selective herbicides to tackle specific weed infestations present in certain regions of their fields, thereby promoting weed control and helping prevent herbicide resistance. As outcomes, it is expected that the prototype developed by the end of this project will achieve detection and localized spraying accuracy above 99% in desiccation applications and at least 95% in post-emergence soybean crop applications, positioning itself as a market leader among competitors and driving Agrio Tecnologia to prominence in the Precision Agriculture sector. (AU)

Articles published in Agência FAPESP Newsletter about the research grant:
Articles published in other media outlets (0 total):
More itemsLess items

Please report errors in scientific publications list using this form.