Transformers for detection

Object Queries

Learnable positional embedding vectors that serve as slots for each potential object prediction, interacting with image features through the attention mechanism to extract relevant information.

← Kembali