An Attention Mechanism for Forest Tree Segmentation based on Efficient Convolutional Neural Networks

It is critical to have accurate semantic categorization of standing trees in order to extract relevant information about standing trees automatically and efficiently from digital photos. The effective categorization of several standing trees set against complicated backdrops is the goal of various classic approaches, however these methods suffer from drawbacks such as poor segmentation accuracy and the need for human intervention. This article presents Modified Convolutional Neural Network,which is based on MobileNet (CNNM), a lightweight network segmented model that is based on deep learning. The goal of this model is to accomplish correct categorization of standing tree photos in an efficient manner. For the purpose of performing multi-scale fusion of the convolutional features of standing trees in images, asemantic segmentation has been selected as the base framework. This decision was made in order to minimize the loss of image edge details that occurs during the standing tree segmentation process, as well as minimize the loss of feature data. MobileNet, which is a lightweight network, has been incorporated into the backbone network in order to minimize the complexity of the computations. The lightweight network segmentation model CNNM based on deep learning that was introduced in this study has the capability of segmenting numerous standing trees with a high degree of precision.