RGB no more: Minimally-decoded JPEG Vision Transformers — mentioned by namjh

Fri, 14 Jul 2023 05:17:38 GMT

Might be related: https://arxiv.org/abs/2211.16421 It's a paper about directly learning via JPEG encodings which works well with visusal transformers' patch mechanism.

Paper and discussion history. A mention is not an endorsement.

RGB no more: Minimally-decoded JPEG Vision Transformers — mentioned by qwertox

Thu, 13 Jul 2023 16:29:27 GMT

Thank you. First published 2022-11-29 on arxiv [0] and updated one month ago. Interesting line: "With these two improvements -- ViT and data augmentation -- we show that our ViT-Ti model achieves up to 39.2% faster training and 17.9% faster inference with no accuracy loss compared to the RGB counterpart." [0] https://arxiv.org/abs/2211.16421v2

Paper and discussion history. A mention is not an endorsement.

RGB no more: Minimally-decoded JPEG Vision Transformers · HackerPapers discussions

RGB no more: Minimally-decoded JPEG Vision Transformers — mentioned by namjh

RGB no more: Minimally-decoded JPEG Vision Transformers — mentioned by qwertox