Skip to content
Aoki Media Sensing Lab.FACULTY OF SCIENCE AND TECHNOLOGYAoki Lab.MEDIA SENSING LAB.
ENJP

Site search

Search by research topic, paper title or author.
HOMEAccepted to ECCV 2026 & Released Code, Pretrained Models, and Dataset (Joint research with ZOZO NEXT)

Accepted to ECCV 2026 & Released Code, Pretrained Models, and Dataset (Joint research with ZOZO NEXT)

The results of joint research with ZOZO NEXT Inc. have been accepted at one of the top international conferences in the field of computer visionECCV 2026(European Conference on Computer Vision).

Paper Title
Reference-Free Image Quality Assessment for Virtual Try-On via Human Feedback

Authors
Yuki Hirakawa, Takashi Wada, Ryotaro Shimizu, Takuya Furusawa, Yuki Saito, Ryosuke Araki, Tianwei Chen, Fan Mo, Yoshimitsu Aoki

Research Personnel
Hirakawa (D1)

Overview
This research proposes a novel image quality assessment (IQA) method that evaluates the quality of virtual try-on (VTON) images in a reference-free manner, without requiring reference images. VTON-IQA In addition, we constructed a large-scale benchmark consisting of 62,700 try-on images and 431,800 human annotations.VTON-QBench We have publicly released the code, pretrained models, and VTON-QBench. An interactive demo is also available on Hugging Face.