Predicting weed invasion in a sugarcane cultivar using multispectral image

ABSTRACT

The cultivation of sugar cane has been gaining great focus in several countries due to its diversity of use. The modernization of agriculture has allowed high productivity, which is affected by the invasion of weeds. With sustainable agriculture, the use of herbicides has been increasingly avoided in society, requiring more effective weed control methods. In this paper, we propose a statistical model capable of identifying the invasion of weeds in the field, using four color spectra as regressor variables obtained by a multispectral camera mounted on an unmanned aerial vehicle. With the exact identification of the weed infestation, it is possible to carry out the management in the field with herbicide applications in the exact places, thus avoiding the increase of the cost of production or even dispensing with the use of herbicides, effecting the mechanical removal of them. Results show that in the experimental field, it was possible to reduce herbicide spraying by 57%.

KEYWORDS:

• GAMLSS
• multinomial logistic regression
• modern agriculture
• statistical modeling
• weed management

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Thiago G. Ramires http://orcid.org/0000-0002-1972-7045

Christel Faes http://orcid.org/0000-0002-1878-9869

Taciana V. Savian http://orcid.org/0000-0002-3309-6075