Discriminating Pennisetum alopecuoides plants in a grazed pasture from unmanned aerial vehicles using object‐based image analysis and random forest classifier

Abstract

Timely and accurate weed detection in pasture is critical for efficient grazing management. Although high‐resolution images from unmanned aerial vehicles (UAVs) offer new opportunities for the detection of weeds at the farm scale, pixel‐based image analyses do not always produce the best results and object‐based image analysis (OBIA) has improved weed discrimination accuracy. In the present study, we evaluated the performance of OBIA on UAV images by integrating random forest (RF) classifier with auxiliary information layers to discriminate and map Pennisetum alopecuroide plants, a prolific and harmful weed, in a grazed pasture. The UAV images were captured at different flight altitudes (28, 56, 82 and 114 m). The OBIA‐RF algorithm included 20 input features: five layers (red‐green‐blue [RGB] or hue‐saturation‐brightness [HSV] image bands, texture and digital surface model) and the descriptive statistics (median, standard deviation, minimum and maximum) for each object. The predicted P. alopecuroides maps were evaluated for out‐of‐bag accuracy and generalized error accuracy in the test dataset. HSV‐based classification had higher classification accuracy, and the lowest altitude of 28 m (spatial resolution, 0.9 cm) was considered the most suitable for the weed detection. Overall, the optimal classification accuracy was achieved in the HSV‐based OBIA‐RF model using the images from the lowest altitude (highest spatial resolution). Among the 20 input features, the brightness information (V layer) in the HSV images was considered the most important because P. alopecuroides ears are black.

Discriminating crops/weeds in an upland rice field from UAV images with the SLIC-RF algorithm

“…In addition, the RF algorithm is receiving increased attention in remote sensing research as a highly suitable approach for high-resolution image data classification (Ma et al, 2015). By combining SLIC and RF, further classification improvements could be obtained (Csillik, 2017;Yasuda, 2018;Yuba et al, 2020b). We initially compared the performance of the SLIC-RF model using different input features of three color spaces (RGB, HSV and L*a*b*) and then combined these features with CHM, Texture and four VIs (ExG, ExR, GRVI, CIVE).…”

“…García-Mateos et al (2015) compared 11 color spaces for classifying soil and plants in lettuce (Lactuca sativa) cultures using images from outdoor fields and found that the L*a*b* color space achieved the best performance with a* channels with 99.2% correct classification. Yuba et al (2020b) compared RGB and HSV color UAV images in discriminating Pennisetum alopecuoides plants in a pasture, and the results showed improved discrimination accuracy in the HSV-based SLIC-RF classifier. In the present study, by assessing the importance of input features (Figures 8 and 9), the HSV feature was considered the most important variable influencing the discrimination of crops, weeds and soils in the early growth stage of upland rice fields.…”

“…Classification accuracy could be enhanced by combining the RF classifier with OBIA (De Castro et al, 2018). In addition, previous studies have reported that weed detection from UAV images could be improved by adding auxiliary information layers, such as spatial texture and vegetation height estimated from UAV digital surface models (DSMs) (Yuba et al, 2020b;Zisi et al, 2018).…”

How Can Unmanned Aerial Vehicles Be Used for Detecting Weeds in Agricultural Fields?

“…The images from Canon S110 NIR (green, red, near-infrared) on UAV were also classified by Reis et al [42] by combining RF and maximum likelihood algorithms. Interestingly, Yuba et al [96] has tried to integrate OBIA and RF with auxiliary information layers in mapping P. alopecuroides, and they found that the combination of these algorithms has increased classification accuracy which is out of bag accuracy = 0.99 and generalized error accuracy i.e., 1.00 from the lowest altitude of 28 m.…”

A Scalable, Supervised Classification of Seabed Sediment Waves Using an Object-Based Image Analysis Approach

“…The few studies that explore this develop algorithms that run over the same region on the same data where image blurring techniques are applied to imitate the coarsening of spatial resolution [37][38][39]. Furthermore, attempts have been made to increase the relative altitude of the sensor to decrease the spatial resolution of the data [85]. Such methods do not effectively capture the relationship between feature's spatial resolution and the scale of measurement.…”