Shenghan Chen | 陈晟晗

I am an undergraduate student majoring in Software Engineering at Shandong University. I am currently working as an Undergraduate Research Assistant at the SDU Time Lab, supervised by Prof. Xiankai Lu and Prof. Yilong Yin. Previously, I also collaborated with Prof. Yawei Li from ETH Zürich.

My research interests primarily lie in Computer Vision, Multi-modal Learning, Vision-Language-Action (VLA) models, and Efficient AI.

Email / CV / CSDN / Github
Shenghan Chen Profile Photo

Publications

* indicates equal contribution.

Reframing Long-tailed learning

Reframing Long-tailed learning via Loss Landscape Geometry

Shenghan Chen*, Yiming Liu*, Yanzhen Wang, Yujia Wang, Xiankai Lu

Accepted by CVPR, 2026

[arXiv] [Code] [Project Page]

We investigate the head-tail seesaw dilemma from the loss landscape view. By transferring long-tailed learning into a continual learning task, this paper uses continuous learning and sharpness perception to improve model attention to tail classes without relying on external synthetic data.

Towards Refined Data Augmentation

Towards Refined Data Augmentation for Long-tailed Recognition

Shenghan Chen, Yiming Liu, Yanzhen Wang, Yujia Wang, Xiankai Lu, Yawei Li

Submitted to Machine Intelligence Research

[arXiv] [Code] [Project Page]

We rethink data augmentation in Long-Tailed recognition, focusing on data distribution shift. Based on large models, we propose automated data augmentation to solve the bias problem caused by data imbalance and the mismatch between synthetic and raw data distributions.

Awards & Honors

Xiaomi Scholarship

Shandong University

2022

National Scholarship

Ministry of Education

2023

Outstanding Graduate of Shandong University

Shandong University

2026