可提示分割、开放词表与密集预测 - Category - 堂堂一跑堂

可提示分割、开放词表与密集预测

2026

TUE-CD：震后建筑变化检测真正难的是短时间隔下的侧视错位 06-15 19:00

CSI-Net：变化检测里真正难的是压住伪变化 06-15 01:00

RSKT-Seg：开放词表遥感分割需要自己的 benchmark 06-14 13:00

Training-Free Text-Based RS Segmentation：让 VLM 只负责选 mask 和点 SAM 06-14 09:00

SST-CD：把无标签建筑变化检测从差异图变成自训练 06-12 17:00

ZODS-RS：DINOv3+SAM2 的零训练遥感检测与实例分割 06-12 15:00

RS-40 Polygon-Native Mask Decoder 06-07 09:39

RS-33 Hyperspectral Foundation Model Transfer 06-07 09:32

RS-28 Mamba/SSM Backbones for Dense Remote Sensing Prediction 06-07 09:27

RS-17 Latent Flow/Diffusion for Change Detection 06-07 09:16

RS-15 Taxonomy-Aware Mask Selection 06-07 09:14

RS-14 SAM3 Presence Score Calibration for Remote Sensing OVSS 06-07 09:13

RS-13 SAM2 Geographic Memory for Multi-Temporal Remote Sensing 06-07 09:12

RS-12 Training-Free Open-Vocabulary Remote Sensing Segmentation 06-07 09:11

RS-11 Reference-Guided SAM for Few-Shot Remote Sensing Segmentation 06-07 09:10

RS-09 HBB/OBB/Mask Unified Visual Grounding 06-07 09:08

RS-01 SAM Box/Coarse-Mask to Point Prompt Refinement for Optical Remote Sensing 06-07 09:00

RSRefSeg：用CLIP+SAM双剑合璧，让遥感图像"听懂"自然语言描述 06-01 23:55

SatDINO：当DINO遇上遥感——对比学习在卫星图像预训练中的深度探索 06-01 22:30

RSKT-Seg：旋转不变+领域迁移实现遥感图像开放词汇分割 06-01 22:30

1
2