Transformers for detection

Bipartite Transformer

Variant of the Transformer architecture where attention mechanisms are applied between image features and a small fixed set of learnable object queries, enabling parallel object prediction.

← Back