Yufan Ren

Hello! I am currently a direct Ph.D. student (2020-) at Image and Visual Representation Lab, part of the School of Computer and Communication Sciences at EPFL, Switzerland. Under the guidance of Prof. Sabine Süsstrunk and Dr. Tong Zhang, I work on 3D vision, Neural Rendering, and Diffusion Models.

During my Ph.D., I did two internships, one NVIDIA Zurich on learning-based robot perception under Dr. Alexander Millane, another at Meta London under Dr. Filippos Kokkinos.

Before my Ph.D. study, I earned my bachelor's degree from Zhejiang University, Hangzhou, where I was honored to receive the Chu Kochen Award.

I am actively seeking opportunities for industry jobs and postdoctoral positions starting in 2025.

Email  /  Google Scholar  /  LinkedIn  /  GitHub

🎉 [Aug 2024] I'm glad to share that I will work as an Intern on the 3D Gen team at Meta London!

🎉 [March 2024] I'm glad to share that starting this March I will work as an Intern on 3D perception at NVIDIA in Zurich with the nvblox team!

🎉 [Oct, 2023]: The next International Conference on Computational Photography (ICCP) is happening at EPFL, Switzerland in 2024. I am excited to announce my role as website chair.

🎉 [Sept 2023]: I am invited for a talk at AWS LauzHack Cloud Research Day on DeepFakes detection.

🎉 [Aprl 2023]: I am admitted to the International Computer Vision Summer School (ICVSS). See you in Sicily 🏝!

Point Cloud Registration (PCR) estimates the relative rigid transformation between two point clouds. We propose formulating PCR as a denoising diffusion probabilistic process, mapping noisy transformations to the ground truth.

We introduce VolRecon, a novel generalizable implicit reconstruction method with Signed Ray Distance Function (SRDF). To reconstruct the scene with fine details and little noise, VolRecon combines projection features aggregated from multi-view features, and volume features interpolated from a coarse global feature volume.

We propose a representational model for image pairs such as consecutive video frames that are related by local pixel displacements, in the hope that the model may shed light on motion perception in primary visual cortex (V1). The model couples the following two components: (1) the vector representations of local contents of images and (2) the matrix representations of local pixel displacements caused by the relative motions between the agent and the objects in the 3D scene.

We introduce BlendedMVS, a novel large-scale dataset, to provide sufficient training ground truth for learning-based MVS. To create the dataset, we apply a 3D reconstruction pipeline to recover high-quality textured meshes from images of well-selected scenes. Then, we render these mesh models to color images and depth maps. To introduce the ambient lighting information during training, the rendered color images are further blended with the input images to generate the training input.

The school aims to provide a stimulating opportunity for young researchers and Ph.D. students. The participants will benefit from direct interaction and discussions with world leaders in Computer Vision. Participants will also have the possibility to present the results of their research, and to interact with their scientific peers, in a friendly and constructive environment.

Peter Grönquist and I did this challenge and won the 100,000 USD prize (incl. grant). In this challenge, we design machine learning models to detect three types of fakeness, i.e., fake faces (DeepFakes), manipulated audio, and mis-synchronization (lip-sync), and use engineering tricks to make it fast.