Development of Weeds Density Evaluation System Based on RGB Sensor

• Published in: IOP conference series. Earth and environmental science Vol. 147; no. 1; pp. 12047 - 12054
• Main Authors: Solahudin, M, Slamet, W, Wahyu, W
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.05.2018
• Subjects: Artificial neural networks; Density; Dosage; Field tests; Herbicides; Neural networks; Nutrients; Soil chemistry; Soil nutrients; Soil water; Weed control; Weeds
• ISSN: 1755-1307; 1755-1315
• DOI: 10.1088/1755-1315/147/1/012047

Weeds are plant competitors which potentially reduce the yields due to competition for sunlight, water and soil nutrients. Recently, for chemical-based weed control, site-specific weed management that accommodates spatial and temporal diversity of weeds attack in determining the appropriate dose of herbicide based on Variable Rate Technology (VRT) is preferable than traditional approach with single dose herbicide application. In such application, determination of the level of weed density is an important task. Several methods have been studied to evaluate the density of weed attack. The objective of this study is to develop a system that is able to evaluate weed density based on RGB (Red, Green, and Blue) sensors. RGB sensor was used to acquire the RGB values of the surface of the field. An artificial neural network (ANN) model was then used for determining the weed density. In this study the ANN model was trained with 280 training data (70%), 60 validation data (15%), and 60 testing data (15%). Based on the field test, using the proposed method the weed density could be evaluated with an accuracy of 83.75%.