I am Dr. Javier
Civera, associate professor at the SLAM Lab, Robotics, Perception and
Real-Time Group, in the University
research interests are within Computer Vision and Robotics,
specifically Structure from Motion, SLAM and visual recognition. I am
currently lecturing Computer Vision, Machine Learning and AI courses at
Bachelor and Master level.
released a dataset for object learning from multimodal human-robot
interaction, as part of the CHISTERA-IGLU project. Find all the details
IGLU, standing for "Interactive Grounded Language Understanding", aims
to robotic knowledge acquisition, modelling and grounding from
multimodal human-robot interaction. I believe that interaction is
needed for small data contexts, and that using deep learning with
spatio-temporal, multimodal data could improve even more its
Our dataset is the first of its class, including audio+RGB-D
synchronized recordings of several users naturally interacting with
objects (pointing, showing, telling their names) from a robocentric
point of view. Our goal is to build visual object models from such
Pablo Azagra, Yoan Mollard, Florian Golemo, Ana Cristina Murillo, Manuel Lopes, Javier Civera
A Multimodal Human-Robot Interaction Dataset (pdf) (poster) (link to the dataset) (video)
Workshop on the Future of Interactive Learning Machines (FILM), together with NIPS 2016 Barcelona, Spain, 2016.
Very happy with our first steps towards fusing deep learning and multiview geometry for monocular mapping!
Fusing depth patterns with the depth estimated from multiple views has
a huge potential to produce accurate and dense reconstructions from a
Find all the details in our arXiv paper and video.
José M. Fácil, Alejo Concha, Luis Montesano, Javier Civera
Deep Single and Direct Multi-View Depth Fusion (arxiv link) (video)
I was invited as a keynote speaker to the International Conference on Cloud and Robotics (ICCR 2016),
in Saint Quentin, France. I had an excellent time with excellent
researchers, discussing the huge potential of the Cloud in our field.
Thank you very much for the invitation, Yulin Zhang!
I am co-editing with Jeannette Bohg, Matei Ciocarlie and Lydia E. Kavraki an Special Issue on Big Data for Robotics, for the Big Data journal. The deadline for submission is June 15th, find the rest of the details below
Jeannette Bohg, Matei Ciocarlie, Javier Civera, Lydia E. Kavraki
Special Issue on Big Data for Robotics
Big Data, Mary Ann Liebert, Inc. publishers
Deadline for paper
submissions: June 15st, 2016.
Full Call for Papers here.
Alejo Concha and myself, in collaboration with Giuseppe Loianno and Vijay Kumar
from UPenn, combined the recent direct monocular SLAM algorithms with
IMU measurements and developed one of the very first Visual-Inertial
Dense SLAM systems. The system runs in real-time at video rate on a
standard CPU. Its main strengths over pure monocular sensing are a
higher degree of robustness and the estimation of the real scale of the
maps and camera motion --both essential for robotics applications.
The paper summarizing this research was accepted at ICRA 2016, find the details there and an illustrative video here.
Alejo Concha, Giuseppe Loianno, Vijay Kumar, Javier Civera
Visual-Inertial Direct SLAM (pdf) (video)
International Conference on Robotics and Automation (ICRA2016), Stockholm, Sweden, 2016.
Our WACV submission, a great work leaded by Wajahat Hussain,
was accepted. We show how learning view-invariant features helps in
retrieval and text spotting when you have small data or missing
training spots. The view-invariance is achieved by leveraging
single-view geometry to rectify image patterns into canonical
fronto-parallel views. It is very interesting to see how 3D cues can
help in pattern discovery and recognition!
UPDATE: We compiled all the material related with this paper in this project page https://sites.google.com/site/smalldatawacv16/, including the Matlab code (https://bitbucket.org/nonsemantic/novel_views/get/master.zip) and the Hotels dataset (https://bitbucket.org/nonsemantic/hotels_dataset_wacv/get/master.zip).
Javier Civera, Luis Montano and Martial Hebert
Dealing with Small Data and Training Blind Spots in the Manhattan World (pdf) (project page) (code) (Hotels dataset)
IEEE Winter Conference on Applications of Computer Vision,
(WACV16), Lake Placid, NY, USA, 2016.
released the code of our DPPTAM system, presented at IROS 2015. DPPTAM
is a dense, piecewise-planar monocular SLAM system that runs in
real-time on a CPU.
Find the paper here, a Youtube video here and the code in github at https://github.com/alejocb/dpptam
I have been invited to the Qualcomm AR Lecture Series in Vienna on Nov. 19th. Eager to discuss about vision research and AR applications with them!
We just released S-PTAM, a feature-based stereo SLAM system. Find the
And an illustrative video in this Youtube link: https://youtu.be/kq9DG5PQ2k8
This is the result of a collaboration with Taihú
Pire, Thomas Fischer,
de Cristóforis and Julio
Jacobo, from the Robotics
and Embedded Systems Lab in the University of Buenos Aires,
Find more details in our IROS15 paper:
Taihú Pire, Thomas
Fischer, Javier Civera, Pablo de Cristóforis, Julio César Jacobo Berlles
Stereo Parallel Tracking
and Mapping for Robot Localization (pdf)
IEEE/RSJ International Conference on Intelligent
Robots and Systems (IROS15), Hamburg, Germany, 2015
3 IROS papers accepted! :D
Together with Giuseppe Serra, Rita
Cucchiara and Kris
Kitani we are organizing a Special
Issue on Wearable and Ego-vision Systems for Augmented Experience,
for the IEEE Transactions on Human-Machine Systems. The relevant info
Giuseppe Serra, Rita Cucchiara, Kris M.
Kitani, Javier Civera
Special Issue on Wearable and Ego-vision Systems for Augmented
Transactions on Human-Machine Systems
submissions are due October 1st, 2015.
The issue will be published
Download the Call for Papers here.
I will be at OCEANS
2015 MTS/IEEE to present our work using direct monocular SLAM
methods on underwater sequences, in collaboration with Paulo Drews-Jr, from Universidade
Federal do Rio Grande in Brazil, and Mario Campos,
from Universidade Federal de Minas Gerais in Brazil.
Alejo Concha, Paulo
Drews-Jr, Mario Campos and Javier Civera
Real-Time Localization and Dense Mapping in
Underwater Environments from a Monocular Sequence (draft)
MTS/IEEE OCEANS, Genova, Italy,
Issue on Cloud Robotics and Automation at IEEE Transactions on
Automation Science and Engineering is out! After a year and a half
of hard work, I am happy with the result and happy to be involved in
such an exciting topic.
I joined the Editorial Board of the IEEE Transactions on Automation
Science as an Associate Editor for the next 3 years :)
From July 28th to August 2nd I will be giving a week-long course on
robotic vision at the Escuela de Ciencias
Informáticas 2014 in Buenos Aires, Argentina. Thanks to the Computer Science department in the
University of Buenos Aires for the invitation!
On June 30th, I am giving an invited talk at MUVS (Multi Unmanned Vehicle
in Compiegne, about robotic vision. The workshop looks extremely
interesting; it is directed both to industry and academia, and it
addresses the vision, control and network challenges of unmanned
vehicles. Can't wait to be there!
happy with our accepted paper at RSS 2014! We basically integrate
Manhattan and piecewise-planar constraints in the variational approach
to monocular mapping. Such constraints are of importance in indoor and
outdoor man-made scenes, where the photometric constraints and standard
regularization can lead to inaccurate results in areas of low texture.
Our approach shows to reduce the errors in a room-sized environment in
a factor 5x.
More details in the paper and the video:
Alejo Concha, Wajahat
Hussain, Luis Montano, Javier Civera
Manhattan and Piecewise-Planar Constraints
for Dense Monocular Mapping (draft) (video)
2014 Robotics: Science and Systems (RSS2014),
Berkeley, USA, 2014.
with Wajahat Hussain and Luis Montano, I have been working on room
layout estimation from audiovisual data. We got a paper accepted at
Javier Civera and Luis Montano
Echoes in Single View Geometry Estimation (pdf)
Twenty-Eighth Conference on Artificial
Intelligence (AAAI-2014), Québec City, Canada,
are quite excited about the possibilities of this new line of research!
Audio is a quite unexplored data format that is present in most
portable devices today (smartphones, kinect, laptops...).
paper with Alejo Concha on the use of superpixels in monocular SLAM was
accepted for ICRA 2014 (UPDATE: And nominated for the Best Vision Paper
Award!!). Find the paper here
Alejo Concha and
in Monocular SLAM (pdf draft)(video)
IEEE International Conference on Robotics and
Automation (ICRA) 2014.
Code to be released soon!
Our proposal for doing SLAM in the Cloud has been accepted in Robotics
and Autonomous Systems.
Luis Riazuelo, Javier
Civera, J. M. M.
C2TAM: A Cloud
Framework for Cooperative Tracking and Mapping (pdf draft)(video 1)(video 2)
Robotics and Autonomous Systems, accepted for
publication in 2013.
Code available at https://sites.google.com/site/c2tamvisualslam/
to be released soon!
Following our Cloud
Robotics Workshop at IROS 2013,
I am co-organizing a special issue on Cloud Robotics and Automation at IEEE
Transactions on Automation Science and Engineering, with Markus Waibel (ETHZ), Matei
Ciocarlie (Willow Garage), Alper
Aydemir (NASA/JPL), Kostas
Bekris (Rutgers University) and Sanjay Sarma (MIT) (and
the invaluable help of IEEE T-ASE Editor-in-Chief Ken Goldberg).
the Call for Papers here.
I am co-organizing a workshop at ICCV
2013 on Wearable Vision, together with my colleagues Ana Cristina Murillo
from my group in the University of Zaragoza and Mohammad Moghimi and Serge Belongie from
University of California San Diego.
I am quite excited to be involved in such a promising
topic; and also to have as invited speakers Kristen Grauman, Takeo Kanade,
Milanfar, Jim Rehg
Thomas. I am sure it will be a great venue!
I am co-organizing a workshop at IROS
2013 on Cloud Robotics, together with my RoboEarth colleague Markus Waibel (ETHZ), Alper Aydemir (KTH), Ken Goldberg (Berkeley) and Matei
Ciocarlie (Willow Garage). Check the workshop web for updates:
I hope to see you there if you go to IROS, it will be an excellent
opportunity to learn and discuss about the great potential of Cloud
Computing for robots!
I am looking for a PhD student on SLAM with a wearable camera, starting
January 2013. Please, click here
for more details on the position and how to apply.
I will be at CVPR 2012 with the paper
Christian Leistner, Javier Civera, Cordelia Schmid, Vittorio Ferrari
Class Detectors from Weakly Annotated Video (pdf)
Thank you very much Alessandro
Schmid and Vitto
Ferrari for the opportunity of working with you, I learned a lot
from you all.
|May 2012. I
gave a SLAM course in the Mechatronics
department at Tallinn
University of Technology. I really enjoyed my time there, thank you
very much Professor Mart Tamre and Dmitry Shvarts for your kind
A year went by so fast!! An extremely busy and exciting year, with many
news but no time to update them here: I had an excellent time during my
4 months visit at Calvin
group, ETHZ. I became an associate professor at Universidad de
Zaragoza, Spain. I coauthored an IJCV paper with Joan Solà,
Vidal-Calleja and J.
M. M. Montiel. My
PhD thesis was published by Springer...
I am organizing, together with Ana Cristina Murillo, Kurt
Konolige and Jana
Kosecka the 1st IEEE Workshop on
Challenges and Opportunities in Robots Perception, held jointly
with ICCV 2011 in Barcelona.
Please, consider submitting a paper if you are working on any topic
related with robotic perception. Deadline is July 1. For more
information, visit our web: http://robots.unizar.es/corp
I am in ETH in Zürich on a research visit until July, working with
My PhD thesis was accepted for publication in the series Springer Tracts in Advanced
I started co-advising the PhD thesis of Mr. Wajahat Hussain together
with Prof. Luis Montano.
I was proud to be a finalist for the Robotdalen Award, one of the
most prestigious international awards for PhD theses in Robotics and
Automation. In the picture I am with Uwe Mettin
(left), now in Trondheim; and Matei
Ciocarlie (right), the winner, now at Willow Garage.
Picture taken by the photographer Terése Andersson.
I released some Matlab code from my PhD thesis under GPL License
--inverse depth and 1-point RANSAC--. Find
it here. I also
news section in my web :)