Abstract:Tree species classification is an important and challenging task in image recognition and the management of forest resources. Moreover, the task of tree species classification based on remote sensing images can significantly improve the efficiency of the tree species survey and save costs. In recent years, many large models have achieved high accuracy in the task of tree species classification in an airborne remote-sensing manner, but due to their fixed geometric structure, traditional convolutional neural networks are inherently limited to the local receptive field and can only provide segmental context information. The limitation of insufficient context information greatly affects the segmentation accuracy. In this paper, a dual-attention residual network (AMDNet) and a re-parameterized model approach are proposed to capture the global context information, fuse the weight, reduce the model volume, and maintain the computational efficiency. Firstly, we propose MobileNetV2 as the backbone network for feature extraction to further improve the feature identification by modeling semantic dependencies in the spatial dimension and channel dimension and adding the output of the two attention modules. Then, the attention perception features are generated by stacking the attention modules, and the in-depth residual attention network is trained using attention residual learning, through which more accurate segmentation results can be obtained. Secondly, we adopt the approach of structure re-parameterization, use a multi-branch topology for training, carry out weighted averaging on multiple trained models, and fuse multiple branch modules into a completely equivalent module in inference. The proposed approach results in a reduction in the number of parameters and an accelerated inference speed while also achieving improved classification accuracy. In addition, the model training strategy is optimized based on Transformer to enhance the accuracy of segmentation. The model was used to conduct classification experiments on aerial orthophotos of Hongya Forest Farm in Sichuan, China, and the MIOU of tree species recognition using the test equipment reached 93.8%. Compared with current models such as UNet, our model exhibits a better performance in terms of both speed and accuracy, in addition to its enhanced deployment capacity, and its speed advantage is more conducive to real-time segmentation, thereby representing a novel approach for the classification of tree species in remote sensing imagery with significant potential for practical applications.

ALResNet: Attention-Driven Lightweight Residual Network for Fast and Accurate Image Recognition

Real-Time Facial Landmark Detection by Attention-driven Lightweight Network

A lightweight and stochastic depth residual attention network for remote sensing scene classification

LWRN: Light-Weight Residual Network for Edge Detection

Improved Residual Networks for Image and Video Recognition

Efficient ResNets: Residual Network Design

An Image Classification Method Based on Adaptive Attention Mechanism and Feature Extraction Network

Efficient Lightweight Attention Network for Face Recognition

Towards efficient and robust face recognition through attention-integrated multi-level CNN

Attention based lightweight asymmetric network for real-time semantic segmentation

Improved Inception-Residual Convolutional Neural Network for Object Recognition

Residual Feature-Reutilization Inception Network

Lightweight and Scale Adaptive Efficient Backbone Network for Recognition

Low-res MobileNet: An efficient lightweight network for low-resolution image classification in resource-constrained scenarios

RESC: REfine the SCore with Adaptive Transformer Head for End-to-end Object Detection

ParaLkResNet: an efficient multi-scale image classification network

AMDNet: A Modern UAV RGB Remote-Sensing Tree Species Image Segmentation Model Based on Dual-Attention Residual and Structure Re-Parameterization

Extract More from Less: Efficient Fine-Grained Visual Recognition in Low-Data Regimes

IremulbNet: Rethinking the inverted residual architecture for image recognition

An Enhanced Low-Resolution Image Recognition Method for Traffic Environments

Iris-LAHNet: a lightweight attention-guided high-resolution network for iris segmentation and localization