Variable-Granularity Tokenization for High-Resolution Object Detection

Khayrul Islam

ViT detectors fix a uniform token grid before any learned stage. A native-resolution aerial detector must then choose between resolving few-pixel objects and staying inside compute…