Vision Transformers for Detection

Cross-Attention Detection

Bidirectional mechanism where object queries interact with image features to simultaneously localize and classify objects.

← Tillbaka