3D Perception Lab


The 3D Perception Lab at the University of Alicante is a group of researchers interested in the intersection of machine learning and computer vision. Our research mission focuses on various aspects of perception often related with mobile robotics in which we exploit 3D data as the main source of information. Some of our research lines include object recognition, semantic segmentation, rigid and non-ridig registration, visual localization and mapping, behavior analysis, and depth estimation. Apart from those general lines we are also highly interested in making those solutions run efficiently by leveraging GPU acceleration using CUDA. Aside from 3D data as our backbone, we are also tied together by our shared vision in the great potential of artificial intelligence, mainly deep learning, which we try to apply and push its limits in every project we work on.

Research Lines


Our main research lines are Machine Learning (Deep Learning specifically), 3D Computer Vision, and GPU Computing. Although our research usually focus on the intersection of those three fields, we are also open to applying our knowledge to cross-disciplinary problems that can be solved by applying artificial intelligence or computer vision techniques. We are also work on the acceleration of algorithms outside that scope that might benefit from a parallel/GPU implementation.

Deep Learning

Artificial intelligence applied to computer vision and robotics (semantic segmentation, depth estimation, scene understanding, object recognition, localization and maping).


3D Computer Vision

Traditional computer vision methods and new challenges for 3D data (rigid registration, non-rigid registration, reconstruction).


GPU Computing

Acceleration of computer vision methods and artificial intelligence pipelines for real-time execution and maximum efficiency.


Selected Publications


2018

2017

2016

2015

2014

Projects


Our team develops its research lines through various funded projects related to computer vision and robotics. Take a look at our cutting-edge research.


Tech4Diet

This project is aimed at improvement the obesity treatment, from a multidisciplinary perspective by using state of the art technology of 3D modeling or virtual reality to study currently unsolved problems related to the adherence treatment. This opens new opportunities to study dietary and nutritional aspects and provides new challenges in technological research.

Portfolio Item Image

Tech4Diet

COMBAHO

The main scientific objective of this project is to promote the health and well-being of society through the design, development and evaluation of an assistant for people with acquired or dependent brain damage to help them face the challenges posed by their illness in their full social integration in both indoor and outdoor environments.

Portfolio Item Image

COMBAHO

SIRMAVED

SIRMAVED is a multisensor system for rehabilitation and interaction of people with motor and cognitive disabilities. The system enables to perform different therapies using multiple modes of interaction (pose, body and hands gestures, voice, touch and gaze position) depending on the type and degree of disability. Through a training process, the system can be customized enabling the definition of patients’ own gestures for each sensor. The system is integrated with a range of applications for rehabilitation.

Portfolio Item Image

SIRMAVED

Research Highlights


  • All
  • Semantic Segmentation
  • Virtual Reality
  • Dataset
  • Depth Estimation
  • Reconstruction
  • Object Recognition
  • GPU Computing

UnrealROX

UnrealROX is an environment built over Unreal Engine 4 which aims to reduce the reality gap when training deep learning architectures by leveraging hyperrealistic indoor scenes that are explored by robot agents which also interact with objects in a visually realistic manner in that simulated world.

Portfolio Item Image
UnrealROX

Virtual Reality

Interactive 3D Object Recognition

In this work, we propose the implementation of a 3D object recognition system which will be optimized to operate under demanding time constraints. The system must be robust so that objects can be recognized properly in poor light conditions and cluttered scenes with significant levels of occlusion.

Portfolio Item Image
Interactive 3D Object Recognition

Object Recognition

Meet Our Team


Avatar Team Member

Jose Garcia-Rodriguez

Full Professor

Avatar Team Member

Miguel Cazorla-Quevedo

Full Professor

Avatar Team Member

Jorge Azorin-Lopez

Full Professor

Avatar Team Member

Andres Fuster-Guillo

Full Professor

Avatar Team Member

Sergio Orts-Escolano

Assistant Professor

Avatar Team Member

Marcelo Saval-Calvo

Assistant Professor


PhD Students

Avatar Team Member

Alberto Garcia-Garcia

PhD Student | 3D Deep Learning

Avatar Team Member

Sergiu Ovidiu Oprea

PhD Student | Predictive Learning

Avatar Team Member

Zuria Bauer Hartwig

PhD Student | Deep Depth Estimation

Avatar Team Member

Victor Villena-Martinez

PhD Student | Non-rigid Reconstruction

Avatar Team Member

Francisco Gomez-Donoso

PhD Student | Object Recognition

Avatar Team Member

Pablo Martinez-Gonzalez

PhD Student | Deep Online Learning

Avatar Team Member

John Castro-Vargas

PhD Student | Deep Reinforcement Learning

Avatar Team Member

Groot

Mascot


BsC/MsC Students

Avatar Team Member

Andres Carpena-Latour

MsC Automation and Robotics | Deep Reinforcement Learning

Avatar Team Member

Alexei Jilinsky

BsC Multimedia Engineering | Virtual Reality

Avatar Team Member

Adrian Frances-Lillo

BsC Multimedia Engineering | Reinforcement Learning

Avatar Team Member

Pablo Lopez-Iborra

BsC Multimedia Engineering | Virtual Reality

Avatar Team Member

Placido Lopez-Avila

BsC Computer Science | Deep Action Recognition

Avatar Team Member

Jordi Amoros

BsC Computer Science | Deep Learning

Avatar Team Member

Carlos Perello-Camacho

BsC Computer Science | Deep Natural Language Processing