Vision Transformers (ViT)

Patch Embedding

Process of converting image patches into fixed-dimensional embedding vectors through linear projection to feed into the Transformer.

← Geri