Researchers have developed Explicit Semantic Anchors (ESA), a novel method for joint change detection and captioning in remote sensing images. ESA translates visual evidence into semantic conditions, including entity presence, temporal appearance variation, and spatial extent, to guide description generation. This approach utilizes a prediction-guided readout and three factor heads to estimate anchor activations, which then inform the generation process through context and lexicon routes. Experiments on LEVIR-MCI and WHU-CDC datasets demonstrate strong performance, validating the effectiveness of supervised visual attributes as explicit conditions for generating change descriptions without requiring additional language annotations. AI
IMPACT Introduces a novel approach for improving the semantic understanding and description of changes in remote sensing imagery.
RANK_REASON Academic paper detailing a new method for computer vision tasks. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXiv
- CatalyzeX
- CORE Recommender
- DagsHub
- European Space Agency
- Explicit Semantic Anchors
- Gotit.pub
- Hugging Face
- Influence Flower
- LEVIR-MCI
- ScienceCast
- WHU-CDC
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →