Tree trunk instance segmentation is crucial for under-canopy unmanned aerial vehicles(UAVs)to autonomously extract standing tree stem attributes.Using cameras as sensors makes these UAVs compact and lightweight,facili...Tree trunk instance segmentation is crucial for under-canopy unmanned aerial vehicles(UAVs)to autonomously extract standing tree stem attributes.Using cameras as sensors makes these UAVs compact and lightweight,facilitating safe and flexible navigation in dense forests.However,their limited onboard computational power makes real-time,image-based tree trunk segmentation challenging,emphasizing the urgent need for lightweight and efficient segmentation models.In this study,we present RT-Trunk,a model specifically designed for real-time tree trunk instance segmentation in complex forest environments.To ensure real-time performance,we selected SparseInst as the base framework.We incorporated ConvNeXt-T as the backbone to enhance feature extraction for tree trunks,thereby improving segmentation accuracy.We further integrate the lightweight convolutional block attention module(CBAM),enabling the model to focus on tree trunk features while suppressing irrelevant information,which leads to additional gains in segmentation accuracy.To enable RT-Trunk to operate effectively under diverse complex forest environments,we constructed a comprehensive dataset for training and testing by combining self-collected data with multiple public datasets covering different locations,seasons,weather conditions,tree species,and levels of forest clutter.Com-pared with the other tree trunk segmentation methods,the RT-Trunk method achieved an average precision of 91.4%and the fastest inference speed of 32.9 frames per second.Overall,the proposed RT-Trunk provides superior trunk segmentation performance that balances speed and accu-racy,making it a promising solution for supporting under-canopy UAVs in the autonomous extraction of standing tree stem attributes.The code for this work is available at https://github.com/NEFU CVRG/RT Trunk.展开更多
Tree trunks detection and their location information are needed to perform effective production and management in forestry and fruit farming.A novel algorithm based on data fusion with a vision camera and a 2D laser s...Tree trunks detection and their location information are needed to perform effective production and management in forestry and fruit farming.A novel algorithm based on data fusion with a vision camera and a 2D laser scanner was developed to detect tree trunks accurately.The transformation was built from a laser coordinate system to an image coordinate system,and the model of a rectangle calibration plate with two inward concave regions was established to implement data alignment between two sensors data.Then,data fusion and decision with Dempster-Shafer theory were achieved through integration of decision level after designing and determining basic probability assignments of regions of interesting(RoIs)for laser and vision data respectively.Tree trunk width was calculated by using laser data to determine basic probability assignments of RoIs of laser data.And a stripping segmentation algorithm was presented to determine basic probability assignments of RoIs of vision data,by calculating the matching level of RoIs like tree trunks.A robot platform was used to acquire data from sensors and to perform the developed tree trunk detection algorithm.Combined calibration tests were conducted to calculate a conversion matrix transforming from the laser coordinate system to the image coordinate system,and then field experiments were carried out in a real pear orchard under sunny and cloudy conditions,with trunk width measurement of 120 trees and 40 images processed by the presented stripping segmentation algorithm.Results showed the algorithm was successful to detect tree trunks and data fusion improved the ability for tree trunk detection.This algorithm could provide a new method for tree trunk detection and accurate production and management in orchards.展开更多
基金supported in part by the National Natural Science Foundation of China(No.31470714 and 61701105).
文摘Tree trunk instance segmentation is crucial for under-canopy unmanned aerial vehicles(UAVs)to autonomously extract standing tree stem attributes.Using cameras as sensors makes these UAVs compact and lightweight,facilitating safe and flexible navigation in dense forests.However,their limited onboard computational power makes real-time,image-based tree trunk segmentation challenging,emphasizing the urgent need for lightweight and efficient segmentation models.In this study,we present RT-Trunk,a model specifically designed for real-time tree trunk instance segmentation in complex forest environments.To ensure real-time performance,we selected SparseInst as the base framework.We incorporated ConvNeXt-T as the backbone to enhance feature extraction for tree trunks,thereby improving segmentation accuracy.We further integrate the lightweight convolutional block attention module(CBAM),enabling the model to focus on tree trunk features while suppressing irrelevant information,which leads to additional gains in segmentation accuracy.To enable RT-Trunk to operate effectively under diverse complex forest environments,we constructed a comprehensive dataset for training and testing by combining self-collected data with multiple public datasets covering different locations,seasons,weather conditions,tree species,and levels of forest clutter.Com-pared with the other tree trunk segmentation methods,the RT-Trunk method achieved an average precision of 91.4%and the fastest inference speed of 32.9 frames per second.Overall,the proposed RT-Trunk provides superior trunk segmentation performance that balances speed and accu-racy,making it a promising solution for supporting under-canopy UAVs in the autonomous extraction of standing tree stem attributes.The code for this work is available at https://github.com/NEFU CVRG/RT Trunk.
基金The study was supported by“Jiangsu Provincial Natural Science Foundation of China(No.BK20151436)Blue Project of Jiangsu Province”.
文摘Tree trunks detection and their location information are needed to perform effective production and management in forestry and fruit farming.A novel algorithm based on data fusion with a vision camera and a 2D laser scanner was developed to detect tree trunks accurately.The transformation was built from a laser coordinate system to an image coordinate system,and the model of a rectangle calibration plate with two inward concave regions was established to implement data alignment between two sensors data.Then,data fusion and decision with Dempster-Shafer theory were achieved through integration of decision level after designing and determining basic probability assignments of regions of interesting(RoIs)for laser and vision data respectively.Tree trunk width was calculated by using laser data to determine basic probability assignments of RoIs of laser data.And a stripping segmentation algorithm was presented to determine basic probability assignments of RoIs of vision data,by calculating the matching level of RoIs like tree trunks.A robot platform was used to acquire data from sensors and to perform the developed tree trunk detection algorithm.Combined calibration tests were conducted to calculate a conversion matrix transforming from the laser coordinate system to the image coordinate system,and then field experiments were carried out in a real pear orchard under sunny and cloudy conditions,with trunk width measurement of 120 trees and 40 images processed by the presented stripping segmentation algorithm.Results showed the algorithm was successful to detect tree trunks and data fusion improved the ability for tree trunk detection.This algorithm could provide a new method for tree trunk detection and accurate production and management in orchards.
