Introduces the UAVReason dataset and UAVReason-Bagel model that jointly train aerial reasoning and cross-modal generation, reporting large gains in VQA accuracy and generation quality on nadir-view UAV imagery.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Can Vision-Language Models Think from the Sky? Unifying UAV Reasoning and Generation
Introduces the UAVReason dataset and UAVReason-Bagel model that jointly train aerial reasoning and cross-modal generation, reporting large gains in VQA accuracy and generation quality on nadir-view UAV imagery.