Automated greening detection (HLB) in the symptomatic and asymptomatic stages in field orange trees: VANT system provided with optical sensors and artificial intelligence

• Grant number: 19/16591-4
• Support type: Research Grants - Innovative Research in Small Business - PIPE
• Duration: April 01, 2020 - March 31, 2022
• Field of knowledge: Agronomical Sciences - Agronomy - Plant Health
• Principal researcher: Adolfo Nicolas Posadas Durand
• Grantee: Adolfo Nicolas Posadas Durand
• Company: Agrientech Limitada
• CNAE: Atividades de apoio à agricultura
• City: São Carlos
• Associated scholarship(s): 21/03016-1 - Monitoring of spectral data acquisition in the field to identify greening in the symptomatic and asymptomatic stage., BP.TT; 20/11448-6 - Spectral data analysis combined with the characterization and processing of multispectral orthomosaics for the detection of greening in the symptomatic and asymptomatic stage, BP.TT; 20/11127-5 - Geoprocessing and remote sensing applied to the construction of a spectral database aiming at the detection of greening in the symptomatic and asymptomatic stage., BP.TT; + associated scholarships 20/11633-8 - Development of unit tests in the acquisition and analysis system and multispectral image classification for greening detection in the symptomatic and asymptomatic stage, BP.TT; 20/11101-6 - Monitoring of the acquisition of spectral data in the field to identify greening in the symptomatic and asymptomatic stage, BP.TT; 20/09076-3 - Automated greening detection (HLB) in the symptomatic and asymptomatic stages in field orange trees: VANT system provided with optical sensors and artificial intelligence, BP.PIPE - associated scholarships

Abstract

Greening also known as (huanglongbing / HLB) attacks all types of citrus and there is no cure for diseased plants. The affected new trees do not produce enough and the adults in production suffer a large premature fall of fruits and deplete over time. The bacterium Candidates Liberibacter asiaticus is currently the main cause of the disease in Brazil present in more than 99% of the diseased plants. In Brazil, control is achieved by eradicating plants with symptoms and by visual inspection with well-trained people. Even so, the incidence in the citrus belt was 18.15% in 2018, which corresponded to an increase of 8.5% in the last year. The accuracy of the visual inspection is about 40%, which means that they may be eradicating healthy plants or failing to eradicate plants with greening, which generate foci of diffusion of the disease. Thus, in this research, we are proposing to automate the greening detection system in the symptomatic and asymptomatic stages in the field. For this, we have an already developed prototype of a drone system equipped with optical sensors and physico-mathematical models based on artificial intelligence for detection at both symptomatic and asymptomatic levels. The prototype was field-tested for symptomatic detection and the preliminary results showed a large detection capacity, with some adjustments to be finalized. For detection at an asymptomatic level we still need more research, which is one of the main objectives of this proposal, as well as the implementation of monitoring large areas with few measurements extracting their characteristics of self-similarity based on the physico-mathematical model of fractal theory and multifractal. The expected results will be a great technological potential of field applications in the fight against pests and diseases of diverse cultures which will have a great impact for our business as a company, since at the moment there are no similar competitors in the market, both national and international. Moreover, this will allow us to generate a bank of spectral characteristics or libraries of different types of diseases in different cultures and varieties. (AU)