DeCap: Decoding CLIP Latents for Zero-Shot Captioning via Text-Only Training.

Wei Li, Linchao Zhu, Longyin Wen, Yi Yang

Browse the full ICLR paper archive.