A paper by Associate Professor Satoshi Iizuka has received the MIRU Nagao Award (Best Paper Award), the top honor at MIRU2026, the Meeting on Image Recognition and Understanding. With approximately 1,000 submissions to MIRU2026, this award recognizes the single best paper selected from among them.
"Improving Image-to-Image Translation via a Rectified Flow Reformulation"
Satoshi Iizuka, Shun Okamoto, Kazuhiro Fukui
The work was highly praised for extending existing regression-based image-to-image translation models into continuous-time transport models through an elegantly simple approach—augmenting the input channels with a noise-corrupted image—demonstrating strong generality and practical impact across a wide range of generative research areas.
Paper:
https://arxiv.org/abs/2603.20186
The paper is also scheduled to be presented at ECCV2026, an international conference to be held in Sweden this September.
ECCV2026: https://eccv.ecva.net/
Award details:
https://miru-committee.github.io/miru2026/author/award/
