Ultra-Wide Area
Object Detection
Each image is processed by a single forward pass without tiling (~0.06s, 4090D, single batchsize)
Each image is processed by a single forward pass without tiling (~0.06s, 4090D, single batchsize)
LEVIRDetNet show strong generalization across diverse scenes, categories, resolutions, and imaging conditions without target-domain training.
LEVIRDet covers 30 parent categories and 159 object types under a unified detection protocol, including 45 aircraft types, 13 vehicle types, and 71 ship types