UAV Based Maize Crop Images Classification as Leaf and Soil Using Color Space and Machine Learning Models

Modern technologies such as Unmanned Aerial Vehicles (UAVs), Artificial Intelligence (AI), and Machine Learning (ML) are revolutionizing precision agriculture. This paper is related to the classification of UAV based maize crop images as soil and leaf using color space and ML models. A standardized method for quantitatively representing and manipulating colors numerically, color space models such as HSV (Hue, Saturation, Value) and Grayscale are fundamental to both image processing and computer vision. In this research work, HSV and Grayscale color spaces are specifically taken into consideration to determine the amount of green color present in an image. Color space models use high-resolution UAV photos of maize crops that have been further subdivided into low-resolution sub-images. Numerical datasets are then generated for two classes of images, namely leaf and soil, which are further used to train ML models such as Logistic Regression with L2 Regularization, Support Vector Machine (SVM), and Naive Bayes (NB). Compared to SVM and NB, Logistic Regression with L2 Regularization achieves 99 percent test accuracy.