.png){kind=icon}
.png){kind=icon}
Return
MAS-LAND: A Multi-Agent System for Landslide detection and rapid response
.png){kind=icon}
OA
AI
A
A
F
M
P
M
S
V
F
DOI：10.1016/j.ijdrr.2026.106263.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• MAS-LAND is a multi-agent framework for post-event landslide detection and rapid response. • The workflow integrates landslide detection, infrastructure exposure assessment, and automated reporting. • SegFormer MIT-b2 achieved the best performance, detecting 603 of 806 landslides in the test dataset. • Building and road exposure are evaluated through RT-DETR extraction and OpenStreetMap data. • The final output is an interactive online map integrating a complete operational report for each detected landslide.
Keywords:
landslide detection
multi-agent systems
SegFormer
emergency response
LLM-based reporting
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
4.5
Papers:
6.0K
Citations:
2.1W
Organization
Cited Papers