The Segment Anything Model (SAM) for remote sensing applications: From zero to one shot
Universidade do Oeste Paulista · Circle Park · +4 more institutions
Abstract
Segmentation is an essential step for remote sensing image processing. This study aims to advance the application of the Segment Anything Model (SAM), an innovative image segmentation model by Meta AI, in the field of remote sensing image analysis. SAM is known for its exceptional generalization capabilities and zero-shot learning, making it a promising approach to processing aerial and orbital images from diverse geographical contexts. Our exploration involved testing SAM across multi-scale datasets using various input prompts, such as bounding boxes, individual points, and text descriptors. To enhance the model’s performance, we implemented a novel automated technique that combines a text-prompt-derived…
Citation impact
- FWCI
- 54.32
- Percentile
- 100%
- References
- 75
Authors
7- LPLucas Prado OscoCorresponding
Universidade do Oeste Paulista
- QWQiusheng Wu
Circle Park, University of Tennessee at Knoxville
- ELEduardo Lopes de Lemos
Universidade Federal de Mato Grosso do Sul
- WNWesley Nunes Gonçalves
Universidade Federal de Mato Grosso do Sul
- APAna Paula Marques Ramos
Universidade Estadual Paulista (Unesp)
Topics & keywords
- Zero (linguistics)
- Shot (pellet)
- Geography
- Computer science
- Cartography
- Remote sensing
- Philosophy
- Linguistics
Funding
- CDCoordenação de Aperfeiçoamento de Pessoal de Nível SuperiorAward: 88881.311850/2018-01
- CNConselho Nacional de Desenvolvimento Científico e TecnológicoAwards: 433783/2018-4, 405997/2021-3, 305296/2022-1, 308481/2022-4, 310517/2020-6
- FDFundação de Apoio ao Desenvolvimento do Ensino, Ciência e Tecnologia do Estado de Mato Grosso do SulAward: 71/009.436/2022