Encoder-Decoder Architecture for Captioning

Learn the fundamental CNN-RNN architecture: a CNN encodes the image, and an RNN or Transformer decoder generates the caption word-by-word.