2026-10-13T16:01:12Z https://www.scidb.cn/oai

10.57760/sciencedb.j00240.0008s 2026-09-03T09:26:23Z

2026-09-03 SDM-SAM2 doi:10.57760/sciencedb.j00240.0008s en This dataset provides the source code and experimental configurations associated with the paper “Semantic Dynamic Boundary Enhancement for Referring Remote Sensing Image Segmentation: SDM-SAM2.” It is released to support the reproduction and verification of the method implementation, model training, performance evaluation, and ablation procedures reported in the paper. SDM-SAM2 addresses referring remote sensing image segmentation by using Segment Anything Model 2 (SAM2) as the segmentation foundation and a BEiT-3 multimodal encoder to predict pixel-level masks of ground objects specified by natural-language expressions.The package contains the implementation of SDM-SAM2, training and evaluation scripts, experimental configurations, environment specifications, and usage documentation. The method contains three main components: a stage-aware text memory module that stores and retrieves historical text states during hierarchical visual encoding; a task-adaptive multi-layer query fusion module that integrates query states from different layers of the mask decoder; and a text-weighted Sobel boundary loss that provides differentiable boundary supervision on continuous mask probability maps. Independent switches and combined configurations are provided for the three components, supporting experiments with the base model, single modules, two-module combinations, and the complete model. Evaluation scripts are designed to calculate the metrics used in the paper, including mean intersection over union (mIoU), overall intersection over union (oIoU), and precision at different intersection-over-union thresholds (Pr@K).The experiments reported in the paper use two public referring remote sensing image segmentation datasets, RefSegRS and RRSIS-D. Due to the original distribution licenses and file sizes, the remote sensing images, referring expressions, segmentation masks, and third-party pretrained weights are not redistributed in this package. Users should obtain the datasets and pretrained weights for RefSegRS, RRSIS-D, SAM2, and BEiT-3 from their original providers and configure the local paths according to the supplied documentation. Third-party open-source components remain subject to their original licenses, while the authors’ original code, configurations, and documentation are governed by the license specified on the Science Data Bank record. referring remote sensing image segmentation; vision-language fusion; Segment Anything Model 2; stage-aware text memory; multi-layer query fusion; Sobel boundary loss WangYifei Liu Fan PUBLIC https://api.github.com/licenses/apache-2.0 dataset Science Data Bank