Agriculture drone spraying fertiliser on the rice fields .
slight damage , severe damage and dead crops [ 24 ]. The correlation between image-extracted NDVI and sheath blight disease severity was R 2 = 0.627 and the accuracy to quantify different levels of the disease in the plots was 63 % [ 25 ]. The accuracy for weed recognition was 0.883 as documented by Huang et al . ( 2018 ) [ 26 ] and the overall accuracy of weed maps obtained from NDVI was higher than 91 % as reported by Stropping et al . ( 2018 ) [ 27 ]. Therefore , UAV-based mapping techniques can gather images and derive data to generate precise P & D & W distribution maps depicting the spots where chemicals are needed .
The P & D & W distribution maps can be used by the spraying drone to apply precise chemical volumes to the spot areas in the field . The use of the Global Positioning System ( GPS ) for measuring distances allows spraying drones to follow the spraying flight paths and flying heights have been programmed according to the P & D & W distribution maps . This map-based variable rate application of chemicals is useful to reduce chemical and labour costs and improve the ability to apply chemicals with greater precision and on a timely basis .
YIELD PREDICTION
The capability to predict crop yield before harvesting is also an important factor , as it enables farm managers to change farming practices throughout the growing season to maximise profit and yield while minimising costs . Recently , remote sensing technology has been widely used for crop yield prediction . NDVI has been widely used in remote sensing applications for monitoring crop conditions in qualitative and quantitative vegetation analysis [ 28 ]. Regression methods have been used to predict crop yields such as generating a regression model to develop direct empirical relationships between the NDVI measurements and crop yield . A UAV equipped with a camera has been used to predict rice crop yields [ 29 , 30 ]. The results indicated that NDVI values and rice yields showed a good correlation with r 2 = 0.75 [ 29 ] and R 2 = 0.75 [ 30 ]. Hence , the yield estimation regression model using NDVI values was capable of predicting rice yields with high accuracy .
53