Precision Agriculture: Application of Segmentation Techniques to Monitor Lemon Tree Health

This paper explores two segmentation techniques used in precision agriculture: segmentation based on color thresholding and segmentation based on neural networks. This research aims to evaluate these two segmentation techniques using RGB photographs captured by UAV to detect lemon trees. Segmentation based on color thresholding is based on defining color ranges to identify and separate objects of interest in the images. On the other hand, neural network-based segmentation uses deep learning algorithms to predict and delineate objects in images more accurately. The study compares the performance, advantages, and disadvantages of both techniques. The color thresholding technique stands out for its simplicity and computational efficiency and is effective in simple cases with well-defined colors. Likewise, approaches based on neural networks offer greater precision and learning capacity, allowing the segmentation of complex objects and multiple classes for its application in precision agriculture.