Tree Seedlings Detection And Counting Using A Deep Learning Algorithm

With the beginning of artificial intelligence(AI),AI technology has been used in various domains of human life,such as agriculture,industry,the military,medicine,and others.It was considerable progress in human society.AI has replaced labor in all walks of life,significantly reduced costs,improved production efficiency,and has become the preferred driving technology for all industries.Artificial Intelligence is playing an essential role in agriculture.Moreover,it can help farmers to optimize their operations,increase yields,and reduce costs.AI can be used to monitor crops,detect diseases and pests,optimize irrigation systems,and even predict weather patterns.Deep learning algorithms are a type of artificial intelligence(AI)that can learn from data without relying on explicit programming to achieve various tasks.Deep learning algorithms have been used for various tasks,including image classification,object detection,natural language processing,and speech recognition.Deep learning algorithms have recently been applied to tree detection and counting in RGB images.Computer vision algorithms can quickly and accurately count the number of trees in a specific area.The main contents of this study are divided as follows:(1)Computer vision technology is a low-cost and efficient method for identifying and counting objects.Compared with the traditional counting trees methods,which are timeconsuming and laborious,automated tree-counting methods have the characteristics of good sustainability,high efficiency,and strong reliability.This study presents a method for detecting and counting tree seedlings in images using a deep-learning algorithm.The deep learning-based counting trees methods have a high economic value and broad application prospects in detecting the type and quantity of tree seedlings.Thus,using the deep learning-based counting trees methods reduce the cost and time of manual tree counting on a large scale and significantly improves the counting accuracy results.In general,training this type of neural network requires a large amount of image data to be able to generalize well to new data.Thus,the initial dataset of this study was built with three types of tree seedlings: Spruce,Black Chokeberries,and Pinus Sylvestris.The data are augmented via several data augmentation methods to improve the accuracy of the detection model and prevent overfitting.(2)Deep learning has been increasingly applied to many agriculture-related applications in recent years,supplementing conventional computer-vision algorithms for counting agricultural objects.Deep learning applications for tree seedlings detection and counting offer a powerful solution to the challenges of precision agriculture,enabling plantations to increase productivity and sustainability while reducing costs and manual labor.Thus,this study built an object detection network based on the relatively new YOLOv5 by deploying the relevant environment and adjusting the current data and training parameters.Three types of tree seedlings were used for training to obtain training weights.(3)Experimental results on three types of tree seedlings showed that our proposed method could effectively identify and count the tree seedling in an image.Specifically,the MAP of the Spruce tree seedlings,Black Chokeberries tree seedlings,and Pinus Sylvestris tree seedlings are89.8%,89.1%,and 95.6%,respectively.The accuracy of the detection model reached 95.10 %on average(98.58 % for Spruce tree seedlings,91.62 % for Black Chokeberries tree seedlings,and 95.11 % for Pinus Sylvestris tree seedlings).The proposed method can provide technical support for the statistical tasks of counting trees.Additionally,in this study,the trained neural network can detect subtle differences in tree shapes and sizes that may not be visible to the human eye,allowing for more accurate counts of trees in a given area.According to the experimental results,the detection method based on the YOLOv5 neural network can accurately identify and count trees in images,even under challenging conditions in the nursery environment.As deep learning algorithms learn from data,their accuracy will continue to improve over time,making them more reliable for tree-counting tasks.Tree counting can have numerous benefits,and it plays a vital role in vast environmental protection applications,forest resource management,crop yield estimation,and agricultural planning.The number of objects is considered an essential factor in various tasks in the agricultural domain.Automated counting can improve farmers’ decisions about yield estimation,disease prevention,stress detection,etc.Moreover,counting in the agriculture domain could provide valuable temporal,spatial and individual information;for example,counting the number of fruits,plant leaves,and flowers,or even trees or plants per se,can aid farmers in predicting yield,systematically monitoring growth and plant health and hence to make better data-driven decisions.Thus,it improves product quality and quantity through better management and cultivation practices,enabling ideal harvesting timing decisions and fast disease detection.