Evaluation of bag-of-features (BoF) technique for weed management in sugarcane production

Abstract

Weeds interfere in agricultural production, causing a reduction in crop yields and quality. The identification of weed species and the level of infestation is very important for the definition of appropriate management strategies. This is especially true for sugarcane, which is widely produced around the world. The present study has sought to develop and evaluate the performance of the Bag-of-Features (BoF) approach for use as a tool to aid decision-making in weed management in sugarcane production. The support vector machine to build a mathematical model of rank consisted of 30553 25x25-pixel images. Statistical analysis demonstrated the efficacy of the proposed method in the identification and classification of crops and weeds, with an accuracy of 71.6% and a Kappa index of 0.43. Moreover, even under conditions of high weed density and large numbers of overlapping and/or occluded leaves, weeds could be distinguished from crops This study clearly shows that the system can provide important subsidies for the formulation of strategies for weed management, especially in sugarcane, for which the timing of weed control is crucial.

Identification of Corn Leaf Diseases Comprising of Blight, Grey Spot and Rust Using DenseNet-201

“…Bag of Features (Santiago et al, 2019) Consequential trajectories with majority of invalid significances and scant (Hiba et al, 2016).…”