Pablo Palafox

I am a Ph.D. candidate at Angela Dai's 3D AI Lab, at the Technical University of Munich.

I received a Master's Degree in Mechanical Engineering from the Technical University of Munich in 2019, as well as a Master's Degree in Automation and Electronics from the Technical University of Madrid, as part of a Double Degree Program. Prior to that, I received my Bachelor's in Automation and Electronics from the Technical University of Madrid.

Email  /  Google Scholar  /  Twitter  /  Github  /  Youtube

profile photo
News

[July 2021] NPMs will appear at ICCV 2021!
[May 2021] Excited to join Facebook Reality Labs as a Research Intern with Tony Tung
[April 2021] NPMs are out on ArXiv
[March 2021] Aljaz's Neural Deformation Graphs accepted at CVPR 2021 (Oral)
[September 2020] Neural Non-Rigid Tracking accepted at NeurIPS 2020

Research

My research interests lie on 3D tracking and reconstruction of non-rigidly deforming objects in dynamic environments.

TransformerFusion: Monocular RGB Scene Reconstruction using Transformers
Aljaz Bozic, Pablo Palafox, Justus Thies, Angela Dai, Matthias Nießner
ArXiv, 2021
paper | video | bibtex

We introduce TransformerFusion, a transformer-based 3D scene reconstruction approach. The input monocular RGB video frames are fused into a volumetric feature representation of the scene by a transformer network that learns to attend to the most relevant image observations, resulting in an accurate online surface reconstruction.

NPMs: Neural Parametric Models for 3D Deformable Shapes
Pablo Palafox, Aljaz Bozic, Justus Thies, Matthias Nießner, Angela Dai
ICCV 2021
paper | video | bibtex

We propose Neural Parametric Models (NPMs), a learned alternative to traditional, parametric 3D models. 4D dynamics are disentangled into latent-space representations of shape and pose, leveraging the flexibility of recent developments in learned implicit functions. Once learned, NPMs enable optimization over the learned spaces to fit to new observations.

Neural Deformation Graphs for Globally-consistent Non-rigid Reconstruction
Aljaz Bozic, Pablo Palafox, Michael Zollhöfer, Justus Thies, Angela Dai, Matthias Nießner
CVPR 2021 (Oral)
paper | video | bibtex

We introduce Neural Deformation Graphs for globally-consistent deformation tracking and 3D reconstruction of non-rigid objects. Specifically, we implicitly model a deformation graph via a deep neural network and empose per-frame viewpoint consistency as well as inter-frame graph and surface consistency constraints in a self-supervised fashion.

Neural Non-Rigid Tracking
Pablo Palafox*, Aljaz Bozic*, Michael Zollhöfer, Angela Dai, Justus Thies, Matthias Nießner
* Denotes Equal Contribution
NeurIPS 2020
paper | video | code | bibtex

We introduce a novel, end-to-end learnable, differentiable non-rigid tracker that enables state-of-the-art non-rigid reconstruction. By enabling gradient back-propagation through a non-rigid as-rigid-as-possible optimization solver, we are able to learn correspondences in an end-to-end manner such that they are optimal for the task of non-rigid tracking.

Teaching
Teaching Assistant, Practical Course: 3D Scanning and Spatial Learning - Summer Term 2021

Teaching Assistant, Practical Course: 3D Scanning and Spatial Learning - Winter Term 2020/21

Teaching Assistant, Practical Course: 3D Scanning and Spatial Learning - Summer Term 2020

Last updated August 2021.


Source code stolen from Jon Barron.