LDS-Net: A Lightweight Dual-Branch Network for Slender Tree Branch Segmentation in Complex Natural Scenes
Open Access
Abstract
Accurate semantic segmentation of slender curvilinear structures, such as tree branches, in complex natural scenes remains challenging. The main difficulties arise from frequent occlusions, ambiguous boundaries, and limited edge computing resources. To address these issues, we propose LDS-Net, a lightweight dual-branch network designed for thin and continuous branch structures. The Detail-Aware Branch uses the Dynamic Snake Convolution (DSConv) to model irregular local geometry, while the Context-Aware Branch uses a Spatial Efficient Separable Pyramid module (SESP) to capture multi-scale context. To improve segmentation under occlusion and boundary ambiguity, we further integrate a Global Topology Transformer module (GTT) and a boundary guidance mechanism (BG). These features are fused via a Pixel-Wise Attention Fusion module (PAF) and optimized using a multi-head compound loss. Experiments on USTD and N-ABSD show that LDS-Net outperforms six representative networks. It also requires only 14.19 G FLOPs, a 24% reduction compared with PIDNet-s. These results suggest that LDS-Net has potential for future deployment on resource-constrained agricultural and ecological monitoring platforms.Keywords
Funding Information
- National Natural Science Foundation of China (32171788)
This publication has 46 references indexed in Scilit:
- DMSF-Net: Synergistic Integration of Deformable Convolution and Attention Gating for Clinically Efficient Medical Image SegmentationArabian Journal for Science and Engineering, 2026
- Heterogeneous attention fusion network with multi-branch decoding for UAV semantic segmentationJournal of King Saud University Computer and Information Sciences, 2026
- Controllable flow matching synthesis for generalizable crack segmentation with a 50k image–mask benchmarkAutomation in Construction, 2026
- Real-time detection and characterization of trunks and upright branches of pear trees for automatic dormant pruningPrecision Agriculture, 2026
- Autonomous robotic pruning in orchards and vineyards: A reviewSmart Agricultural Technology, 2025
- Enhanced multi-scale detection of olive tree crowns in UAV orthophotos using a deep learning architectureSmart Agricultural Technology, 2025
- Optimizing AlexNet for accurate tree species classification via multi-branch architecture and mixed-domain attentionComplex & Intelligent Systems, 2025
- Tree Branch Characterisation from Point Clouds: a Comprehensive ReviewCurrent Forestry Reports, 2024
- Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation2018
- U-Net: Convolutional Networks for Biomedical Image Segmentation2015