RESEARCH OUTPUT

Publications

8 records2 main-track papers3 workshop honors
01
Grid-Sampler architecture and token pruning overview

See What Matters: Differentiable Grid Sample Pruning for Generalizable Vision-Language-Action Model

Yixu Feng, Zinan Zhao, Yanxiang Ma, Chenghao Xia, Chengbin Du, Yunke Wang, Chang Xu

ICML 2026 CCF-A · CORE A* · First author · Accepted

A plug-and-play differentiable grid sampler for VLA models. With only 16 visual tokens it reduces VLA FLOPs by about 76%, while improving LIBERO success and real-robot performance.

Read paper ↗ · Paper page ↗
02
Copper-Policy compact world representation and action generation overview

Copper-Policy: Focus on the Representation for Robust Robot Manipulation

Zexin Feng, Yixu Feng, Lingyu Xiao, Shang Su, Kexin Zheng, Chang Xu, Mengkai Shi, Shuo Feng, Xintao Yan

arXiv submission · 2026 · Second author · Submitted

Jointly learns a compact world representation and robot policy through temporal joint-embedding prediction, improving robust manipulation without pixel-level future generation.

Read paper ↗ · Paper page ↗
03
HVI-CIDNet architecture from the paper

HVI: A New Color Space for Low-light Image Enhancement

Qingsen Yan*, Yixu Feng*, Cheng Zhang, Guansong Pang, Kangbiao Shi, Peng Wu, Wei Dong, Jinqiu Sun, Yanning Zhang

CVPR 2025 CCF-A · CORE A* · Co-first author · Student first author

Introduces the HVI color space and CIDNet, a lightweight color-intensity decoupling network for robust low-light enhancement.

Paper page ↗ · arXiv ↗
04
FusionNet multi-model fusion framework

FusionNet: Multi-model Linear Fusion Framework for Low-light Image Enhancement

Kangbiao Shi*, Yixu Feng*, Tao Hu, Yu Cao, Peng Wu, Yijin Liang, Yanning Zhang, Qingsen Yan

CVPR Workshops 2025 (NTIRE) · Co-first author · NTIRE 2025 champion

A training-free linear fusion framework combining complementary enhancement models; ranked first in the CVPR 2025 NTIRE low-light enhancement track.

Read paper ↗
05
DiffLight content and detail enhancement framework

DiffLight: Integrating Content and Detail for Low-light Image Enhancement

Yixu Feng, Shuo Hou, Haotian Lin, Yu Zhu, Peng Wu, Wei Dong, Jinqiu Sun, Qingsen Yan, Yanning Zhang

CVPR Workshops 2024 (NTIRE) · First author · NTIRE 2024 · Oral · 4th place

A dual-branch enhancement pipeline with progressive patch fusion for preserving details and reducing block artifacts in UHD images.

Read paper ↗
06
Flow-guided burst HDR restoration framework

Flow-Guided Deformable Alignment with Channel-wise Self-Attention Reconstruct for Efficient Burst HDR Restoration

Weiyu Zhou, Tao Hu, Yixu Feng, Duwei Dai, Yu Cao, Peng Wu, Wei Dong, Yanning Zhang, Qingsen Yan

CVPR Workshops 2025 (NTIRE) · NTIRE 2025 · 2nd place · Burst HDR restoration

An alignment-centric burst HDR restoration model that improves motion alignment while keeping feature fusion lightweight and efficient.

Read paper ↗
07
HVI-CIDNet+ framework for extreme low-light enhancement

HVI-CIDNet+: Beyond Extreme Darkness for Low-Light Image Enhancement

Kangbiao Shi, Xiaowen Ma, Yixu Feng, Tao Hu, Peng Wu, Guansong Pang, Qingsen Yan

IEEE Transactions on Circuits and Systems for Video Technology (TCSVT) CCF-B · Co-author · Journal paper

Extends HVI-CIDNet with a stronger enhancement pipeline for severely underexposed scenes.

Read paper ↗
08
Motivation and color-space comparison from the paper

You Only Need One Color Space: An Efficient Network for Low-light Image Enhancement

Yixu Feng, Cheng Zhang, Pei Wang, Peng Wu, Qingsen Yan, Yanning Zhang

arXiv preprint · First author · 2024

Introduces a trainable HVI color space and CIDNet to stabilize low-light enhancement under noisy illumination conditions.

Read paper ↗