• I will be serving as an Area Chair for NeurIPS 2025.

• We propose FDS, a text-driven image editing method based on frequency-domain decomposition, which enables fine-grained control over local frequency components and extends naturally to 3D texture editing. This work has been accepted to CVPR 2025. Congrats to Yufan,

• Our work UMGen has been accepted to CVPR 2025! UMGen is a unified generative framework that simultaneously predicts and generates key elements of driving scenes—including ego-vehicle actions, maps, traffic participants, and images—enabling both free-form imagination and user-controlled scenario synthesis. Congrats to Yanhao.

• We propose an unsupervised method to recover CLIP's spatial awareness, which has been accepted to ICLR 2025. Congrats to Congpei !

• Our paper AdaNCA, enhancing the robustness of VIT by NCA,  is accepted by NeurIPS 2024. Congrats to Yitao!

• We have 2 papers accepted by ECCV 2024. Congrats to Bahar and Haoqi.

• Our paper, “SINDER: Repairing the Singular Defects of DINOv2", has been selected as an oral presentation at ECCV24. In this work, we introduce a novel approach to examining defects in vision foundation models, achieving significant efficiency improvements—saving over 3,000 GPU hours compared to existing methods. Congratulations to Haoqi on her first project at EPFL! Check our code here.

• Our codes of CVPR 2024 can be found: 360 scene inpainting (InNeRF360), Indoor scene point clouds SSL (OESSL), and relative pose estimation (DVMNet).

• Our MeshNCA is accepted by SIGGRAPH 2024. Feel free to play the demo.

• Congratulations to Dongqing for winning the Rising Star Award at SIGGRAPH 2024! Very well deserved!

• We have 3 papers accepted by CVPR 2024.  Congrats to Chen, Dongqing, and Yanhao. 

• The official code of "Mind Your Augmentation: The Key to Decoupling Dense Self-Supervised Learning" ICLR 24 can be found here.

• The official code of "3D-Aware hypothesis & verification for generalizable relative object pose estimation" ICLR 24 can be found here.

• We have 2 papers accepted by ICLR 2024. Congrats to Congpei and Chen.

• We just open-sourced our paper STSSL (Point clouds SSL), https://github.com/YanhaoWu/STSSL, check it out!

• Our transparent object reconstruction NEMTO is accepted to ICCV 2023. Congrats to Dongqing on her first paper!

• Our VolRecon code can be found here, https://github.com/IVRL/VolRecon/.

• We just released the code of our CVPR 2023 paper DyNCA, https://github.com/IVRL/DyNCA, which generates texture video in real-time.

• We have 4 papers accepted by CVPR 2023, congrats to Yufan, Ehsan, Yanhao, and Bahar on their first CVPR paper.

• Our paper "Modeling Object Dissimilarity for Deep Saliency" was accepted by TMLR, congrats to Bahar and Deblina

• One paper was accepted by ECCV 2022, congrats Di! Multi-view Stereo with NeRF.

• Two papers on representation learning (self-supervised and multi-tasks) were accepted by CVPR 2022

• One paper was accepted by ICASSP 2022

• Our active learning salient object detection was accepted by Pattern Recognition

• Two papers, one oral and one poster, are accepted by CVPR 2021

• Our paper "Learning Saliency from Single Noisy Labelling: A Robust Model Fitting Perspective" is accepted by TPAMI.

• I was awarded the 2019 Chinese Government Award for Outstanding Self-financed Students Abroad.

• Our paper " UC-Net: Uncertainty Inspired RGB-D Saliency Detection via Conditional Variational Autoencoders"   has been nominated for the best paper award in CVPR 2020.

• I will be joining EPFL on October 1st.

• I am fortunate to work with these talented students:

Haoqi Wang, Ph.D. candidate at EPFL 2023.10 ~ now

Yitao Xu, Ph.D. candidate at EPFL 2022.05 ~ now

Chen Zhao,                                     Ph.D. candidate at EPFL 2023.05 ~ now

Yufan Ren, Ph.D. candidate at EPFL 2022.03 ~ now

Dongqing Wang, Ph.D. candidate at EPFL 2022.03 ~ now

Ehsan Pajouheshgar, Ph.D. candidate at EPFL 2021.01 ~ now

Congpei Qiu, Ph.D. candidate at XJTU  2024.05 ~ now (master student 2021.05 - 2024.05)

Yanhao Wu, Ph.D. candidate at XJTU 2023.05 ~ now (master student 2022.05 - 2023.05)

Baran Ozaydin, Ph.D. candidate at EPFL 2021.01 ~ 2024.06

Bahar Aydemir,  Ph.D. candidate at EPFL 2021.01 ~ 2024.06

Deblina Bhattacharjee         Ph.D. candidate at EPFL 2021.01 ~ 2022.09

I've been a committee member (reviewer) for various top-tier journals (TPAMI, TIP, IEEE Signal Processing Magazine, TMLR), and conferences (CVPR, NeurIPS, ICML, ICLR, ECCV, ICCV, AAAI, ACCV, and WACV).