A Novel Change Detection Method Based on Visual Language From High-Resolution Remote Sensing Images
Recently, the release of “all-in-one” foundation models has sparked rapid developments in artificial intelligence. However, due to the fact that these models are typically trained on natural images, their potential in remote sensing remains largely untapped. To address this gap...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10818767/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832540566348365824 |
|---|---|
| author | Junlong Qiu Wei Liu Hui Zhang Erzhu Li Lianpeng Zhang Xing Li |
| author_facet | Junlong Qiu Wei Liu Hui Zhang Erzhu Li Lianpeng Zhang Xing Li |
| author_sort | Junlong Qiu |
| collection | DOAJ |
| description | Recently, the release of “all-in-one” foundation models has sparked rapid developments in artificial intelligence. However, due to the fact that these models are typically trained on natural images, their potential in remote sensing remains largely untapped. To address this gap, this article proposes a novel change detection method based on visual language from high-resolution remote sensing images, named VLCD. Specifically, on the text side, we use context optimization to align text–image semantics. On the image side, we construct a side fusion network, which integrates universal features from the foundation model with domain-specific features from remote sensing through a bridging module. In addition, we introduce a change feature computation module to integrate global features, difference features, and textual information. To validate the effectiveness of the proposed method, we conducted comparative experiments on three public datasets. The results show that the proposed VLCD achieved state-of-the-art <italic>F</italic>1-scores and IoUs on these three datasets: LEVIR-CD (90.99%, 83.46%), SYSU-CD (83.05%, 71.01%), and S2Looking (62.75%, 45.89%), outperforming the results obtained through full fine-tuning while using less than one-tenth of the number of parameters. |
| format | Article |
| id | doaj-art-e0d746c3bba34f2a9743e759dcf717fe |
| institution | Kabale University |
| issn | 1939-1404 2151-1535 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj-art-e0d746c3bba34f2a9743e759dcf717fe2025-02-05T00:00:24ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-01184554456710.1109/JSTARS.2024.352438210818767A Novel Change Detection Method Based on Visual Language From High-Resolution Remote Sensing ImagesJunlong Qiu0https://orcid.org/0009-0001-2139-4550Wei Liu1https://orcid.org/0000-0001-8808-7961Hui Zhang2https://orcid.org/0009-0001-2138-2656Erzhu Li3https://orcid.org/0000-0002-5881-618XLianpeng Zhang4https://orcid.org/0000-0001-9765-9730Xing Li5https://orcid.org/0000-0003-3793-1581School of Geography, Geomatics, and Planning, Jiangsu Normal University, Xuzhou, ChinaSchool of Geography, Geomatics, and Planning, Jiangsu Normal University, Xuzhou, ChinaSchool of Geography, Geomatics, and Planning, Jiangsu Normal University, Xuzhou, ChinaSchool of Geography, Geomatics, and Planning, Jiangsu Normal University, Xuzhou, ChinaSchool of Geography, Geomatics, and Planning, Jiangsu Normal University, Xuzhou, ChinaSchool of Geography, Geomatics, and Planning, Jiangsu Normal University, Xuzhou, ChinaRecently, the release of “all-in-one” foundation models has sparked rapid developments in artificial intelligence. However, due to the fact that these models are typically trained on natural images, their potential in remote sensing remains largely untapped. To address this gap, this article proposes a novel change detection method based on visual language from high-resolution remote sensing images, named VLCD. Specifically, on the text side, we use context optimization to align text–image semantics. On the image side, we construct a side fusion network, which integrates universal features from the foundation model with domain-specific features from remote sensing through a bridging module. In addition, we introduce a change feature computation module to integrate global features, difference features, and textual information. To validate the effectiveness of the proposed method, we conducted comparative experiments on three public datasets. The results show that the proposed VLCD achieved state-of-the-art <italic>F</italic>1-scores and IoUs on these three datasets: LEVIR-CD (90.99%, 83.46%), SYSU-CD (83.05%, 71.01%), and S2Looking (62.75%, 45.89%), outperforming the results obtained through full fine-tuning while using less than one-tenth of the number of parameters.https://ieeexplore.ieee.org/document/10818767/Change detectionfoundation modelprompt learningremote sensing (RS) |
| spellingShingle | Junlong Qiu Wei Liu Hui Zhang Erzhu Li Lianpeng Zhang Xing Li A Novel Change Detection Method Based on Visual Language From High-Resolution Remote Sensing Images IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Change detection foundation model prompt learning remote sensing (RS) |
| title | A Novel Change Detection Method Based on Visual Language From High-Resolution Remote Sensing Images |
| title_full | A Novel Change Detection Method Based on Visual Language From High-Resolution Remote Sensing Images |
| title_fullStr | A Novel Change Detection Method Based on Visual Language From High-Resolution Remote Sensing Images |
| title_full_unstemmed | A Novel Change Detection Method Based on Visual Language From High-Resolution Remote Sensing Images |
| title_short | A Novel Change Detection Method Based on Visual Language From High-Resolution Remote Sensing Images |
| title_sort | novel change detection method based on visual language from high resolution remote sensing images |
| topic | Change detection foundation model prompt learning remote sensing (RS) |
| url | https://ieeexplore.ieee.org/document/10818767/ |
| work_keys_str_mv | AT junlongqiu anovelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT weiliu anovelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT huizhang anovelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT erzhuli anovelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT lianpengzhang anovelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT xingli anovelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT junlongqiu novelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT weiliu novelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT huizhang novelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT erzhuli novelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT lianpengzhang novelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages AT xingli novelchangedetectionmethodbasedonvisuallanguagefromhighresolutionremotesensingimages |