Get Started
Home
Topics
Search
Library
Research questionHow can RGB-to-IR translation improve vehicle detection across unseen UAV domains when target infrared data is scarce?RGB aerial data are more available than thermal imagery, but thermal cues such as engine heat cannot be directly observed in RGB. Translation must therefore bridge the cross-modal gap without losing vehicle features needed for detection across domains.
AI
Computer Vision
Diffusion Models
Evaluation & Benchmarks
Image & Video Processing
Machine Learning
Technology
Latest papersRecent research connected to this question, newest first.RGB-to-IR image translation for infrared vehicle detection in unseen UAV domainsThe evidence uses translators trained on paired source RGB-IR datasets and applied to RGB images from held-out target datasets to generate synthetic IR for vehicle-detector training. Results span five aerial datasets, with Kust4K and VTUAV as target domains; synthetic IR improves over RGB, grayscale, and source-IR baselines, but remains below training with real target-domain IR.research paper · Sep 2, 2026
Related questions
How can object detectors trained on uncamouflaged imagery generalize to unseen camouflage with little target-domain data?How can infrared-visible object detectors remain reliable when one sensor modality is missing?How can aerial visual place recognition adapt across missions without catastrophic forgetting?How can multiple aerial robots identify and track individual wildlife using onboard RGB cameras under decentralized, low-bandwidth coordination?