5 Papers Accepted at ICCV 2025

Congratulations to the following scientists from A*STAR Centre for Frontier AI Research (A*STAR CFAR) on having their papers accepted at the International Conference on Computer Vision (ICCV) 2025:

Prof Ivor Tsang, Director, Distinguished Principal Scientist
Prof Ong Yew Soon, Chief Artificial Intelligence (AI) Scientist and Advisor
Dr Cheston Tan, Senior Principal Scientist
Dr Basura Fernando, Principal Scientist
Dr Li Chen, Senior Scientist
Dr Zhang Hao, Senior Scientist
Ms Yang Hong, Senior Research Engineer
Mr Eric Peh, Research Engineer

List of accepted papers:

1.	GroundFlow: A Plug-in Module for Temporal Reasoning on 3D Point Cloud Sequential Grounding Zijun Lin, Shuting He, Cheston Tan**, Bihan Wen GroundFlow is a new module that helps AI models better follow step-by-step instructions in the task of object grounding in 3D point clouds, by understanding context over time, significantly improving accuracy.
2.	IMoRe: Implicit Program-Guided Reasoning for Human Motion Q&A Chen Li, Chinthani Sugandhika , Yeo Keat Ee, Eric Peh, Hao Zhang, Hong Yang, Deepu Rajan, Basura Fernando* Existing human motion Q&A methods rely on explicit program execution, where the requirement for manually defined functional modules may limit the scalability and adaptability. To overcome this, we propose an implicit program-guided motion reasoning (IMoRe) framework that unifies reasoning across multiple query types without manually designed modules.
3.	Balanced Image Stylisation with Style Matching Score Yuxin Jiang, Liming Jiang, Shuai Yang, Jia-Wei Liu, Ivor W. Tsang**, Mike Zheng Shou Style Matching Score (SMS) is proposed for diffusion-based image stylisation optimisation. SMS matches target style distributions via style-dependent LoRAs, employs Progressive Spectrum Regularisation in the frequency domain, and Semantic-Aware Gradient Refinement for selective stylisation. SMS achieves superior balance between style transfer and content preservation.
4.	Confound from All Sides, Distill with Resilience: Multi-Objective Adversarial Paths to Zero-Shot Robustness Junhao Dong, Jiao Liu, Xinghua Qu, Yew-Soon Ong** The paper introduces a multi-objective adversarial distillation method that generates diverse and disruptive adversaries to better transfer both in-distribution and out-of-distribution robustness from large-scale Vision-Language Models to lightweight ones, ensuring improved zero-shot robust generalisation.
5.	Robustifying Zero-Shot Vision Language Models by Subspaces Alignment Junhao Dong, Piotr Koniusz, Liaoyuan Feng, Yifei Zhang, Hao Zhu, Weiming Liu, Xinghua Qu, Yew-Soon Ong** The paper proposes a novel adversarial fine-tuning framework for Vision-Language Models that aligns subspaces (rather than individual samples) constructed from clean and adversarial image-text pairs, significantly improving generalisable zero-shot robustness across datasets.

*denotes former CFAR student
** denotes current CFAR student

Taking place from 19 – 23 October 2025 at the Honolulu Convention Centre in Hawaii, the International Conference on Computer Vision (ICCV) is a premier event focusing on advances in computer vision, including topics such as image processing, recognition and 3D vision.

Learn more about ICCV 2025.