Valter Piedade

valter.piedade@tecnico.ulisboa.pt


PhD student at Instituto Superior Técnico (IST) since 2022, working at the Institute for Systems and Robotics (ISR-Lisboa). My advisors are Dr. Pedro Miraldo and Prof. José Gaspar.

I have a BSc and a MSc degree in Electrical and Computer Engineering from IST. My major and minor specializations are Systems, Decision and Control and Computers, respectively.

My research interests focus on computer vision, artificial intelligence, and robotics. I work primarily in 3D reconstruction and robust estimation, namely Simultaneous Localization and Mapping (SLAM) systems.


CV
           
profile photo

Research and Projects

Projects:

STORESLAM: Accurate Agent Localization and Mapping Methods for Structured Indoor Retail Store Environments

Visual SLAM system for an autonomous mobile robot in a retail store environment using a single fisheye stereo camera. The system is being designed and implemented from scratch in C++.

Publications:

BANSAC: A dynamic BAyesian Network for adaptive SAmple Consensus
Valter Piedade, Pedro Miraldo
IEEE/CVF International Conference on Computer Vision (ICCV), 2023
Project page - Paper - arXiv - Video - Poster

BANSAC is a sampling strategy for RANSAC. We derive a dynamic Bayesian network that iteratively updates individual data points' inlier probability. In each iteration, the updated probabilities guide the sampling process. This approach outperforms the best baselines in accuracy, being also more efficient.

A Probability-guided Sampler for Neural Implicit Surface Rendering
G. D. Pais, V. Piedade, M. Chatterjee, M. Greiff, and P. Miraldo
European Conference on Computer Vision (ECCV), 2024
Project page - Paper - Video - Poster

Novel pixel sampler for Neural Implicit Rendering that leverages the scene's Signed Distance Function (SDF). Our approach achieves sharper reconstructions and improved performance in less-observed regions, without requiring additional prior data.


Last updated on October 20, 2024. Website source code from Jon Barron.