José Oramas M.

Contact

Office: S.BC.702
University of Antwerp, imec-IDLab
Stadscampus, Sint-Pietersvliet 7
2000 Antwerp
Belgium.

jose.oramas AT uantwerpen DOT be

I come from that little country in the middle of the world, Ecuador. In October 2019, I joined the Internet Data Lab (imec-IDLab) at UAntwerpen as Assistant Professor (tenure track). I received my PhD at the Center for Processing Speech and Images (ESAT-PSI) of KU. Leuven under the advice of Prof. Tinne Tuytelaars and Prof. Luc de Raedt in April 2015. Earlier I received my engineering degree from Escuela Superior Politecnica del Litoral in Ecuador. During my Ph.D. I conducted research on understanding how groups of elements from the image (objects, object-parts, image regions, trajectories, etc.) interact and how the relationships between them can be exploited to improve artificial visual perception problems. This fueled my interest towards investigating exploratory/explanatory models that can identify informative intermediate representations and use them as means to justify the predictions that they make.

A copy of my CV can be found here.

Research Interests

Representation Learning, Model Interpretability/Explainability, Multiple Instance Learning, Computer Vision.

I thank the following institutions for supporting my research:

News

2020-05 : One paper accepted at ICIP 2020 (Preprint).
2019-12 : Our FWO project proposal on the topic of Explainable AI got awarded.
2019-10 : I have joined the IDLab at UAntwerpen as Assistant Professor.
2019-06 : Honored to be considered an Outstanding Reviewer for CVPR 2019.
2019-05 : One paper accepted at ICIP 2019.
2019-03 : I will be giving a tutorial at ECVP 2019.
2018-12 : One paper accepted at ICLR 2019 (Preprint).
2018-10 : Our interpretation/explanation work received a Bell Labs Student Award.
2018-04 : Just received a GPU gift from NVIDIA.
2018-03 : Got among the top-30% scored reviewers at NIPS 2018.
2018-01 : Two papers accepted for WACV 2018.
2017-04 : One paper accepted at CVIU (Preprint).
2017-02 : Just received a generous academic hardware grant from NVIDIA.
2016-09 : I will be attending the Google Computer Vision Summit in Zurich.
2016-08 : One paper accepted at CVIU (Preprint).
2016-07 : I got awarded a KU Leuven PDM grant.
2016-04 : One paper accepted at TPAMI (Preprint).

Teaching Activities

Courses
- Fall 2018, H01D4A : (PO-3) Problem Solving and Engineering Design.
- Fall 2017, H01D4A : (PO-3) Problem Solving and Engineering Design.(demo)
- Fall 2016, H01D4A : (PO-3) Problem Solving and Engineering Design.(demo)
- Fall 2015, H01D4A : (PO-3) Problem Solving and Engineering Design.(demo)
- Fall 2014, H01D4A : (PO-3) Problem Solving and Engineering Design (*).(demo)
- Spring 2014, H09J2A : Pattern Recognition and Image Interpretation.
- Fall 2013, H01D4A : (PO-3) Problem Solving and Engineering Design.(demo)
- Fall 2012, H01D4A : (PO-3) Problem Solving and Engineering Design. (demo)

Doctoral Students
- Kaili Wang (co-supervised with Tinne Tuytelaars).

Master Students
- Roger Granda
- Zehao Wang (co-supervised with Kaili Wang).
- Zhou Wang
- Lies Bollens
- Kevin Bardool
- Shitong Sun
- Jean-Baptiste Verroken
- Chirstiaan Vanbergen
- Kaili Wang (co-supervised with Yu-Hui Huang).
- Seppe Iven (co-supervised with Bert De Brabandere).
- Jonas Heylen (co-supervised with Bert De Brabandere).
- Pieter De Clerq
- Michiel Vanuytsel (co-supervised with Konstantinos Rematas).
- Marc Martínez-Camarena.

Bachelor Students
- Anneline Daggelinckx

Publications

2020

Unpaired Image Shape Translation Across Fashion Data.

Kaili Wang, Liqian Ma, José Oramas M., Luc Van Gool and Tinne Tuytelaars
ICIP 2020 (To appear)
PDF (Preprint) | Supp. Material | Bibtex

Information Compensation for Deep Conditional Generative Networks.

Zehao Wang, Kaili Wang, Tinne Tuytelaars and José Oramas M..
Technical Report
PDF (arXiv:2001.08559) | Bibtex

2019

A Simple Baseline for Multiple Instance and Weakly Supervised Learning

Kaili Wang, José Oramas M., and Tinne Tuytelaars.
Technical Report
PDF (arXiv:1909.05690) | Bibtex

A Systematic Analysis of a Context Aware Deep Learning Architecture for Object Detection.

Kevin Bardool, Tinne Tuytelaars and José Oramas M..
BNAIC/BeNeLearn 2019.
PDF | Ext.Abstract | Bibtex

Visual Explanation by Interpretation: Improving Visual Feedback Capabilities of Deep Neural Networks.

José Oramas M., Kaili Wang, and Tinne Tuytelaars.
ICLR 2019
(Bell Labs Student Award)
PDF (arXiv:1712.06302) | Project website | Bibtex
Future X Days'18 Poster

Towards Object Shape Translation Through Unsupervised Generative Deep Models.

Lies Bollens, Tinne Tuytelaars and José Oramas M.
ICIP 2019 (oral)
PDF | Slides | Bibtex

Detecting intentional self-harm on Instagram: Development, test, and validation of an automatic image recognition algorithm to discover cutting-related posts.

Sebastian Scherr, Florian Arendt, Thomas Frissen, and José Oramas M.
Social Science Computer Review 2019
Link | Bibtex

2018

Unsupervised shape transformer for image translation and cross-domain retrieval.

Kaili Wang, Liqian Ma, José Oramas M., Luc Van Gool and Tinne Tuytelaars
Technical Report
PDF (arXiv:1812.02134)

An Analysis of Human-centered Geolocation.

Kaili Wang, Yu-Hui Huang, José Oramas M., Luc Van Gool, and Tinne Tuytelaars.
WACV 2018
PDF (arXiv:1707.02905) | Project website | Poster | Slides | Bibtex

From Pixels to Actions: Learning to Drive a Car with Deep Neural Networks.

Jonas Heylen, Seppe Iven, Bert de Brabandere, José Oramas M., Luc Van Gool, and Tinne Tuytelaars.
WACV 2018
PDF | Project website | Bibtex

2017

Context-based Object Viewpoint Estimation: A 2D Relational Approach.

José Oramas M., Luc De Raedt and Tinne Tuytelaars.
CVIU 2017.
Link | PDF (arXiv:1704.06610) | Bibtex

Rank Pooling For Action Recognition.

Basura Fernando, Efstratios Gavves, José Oramas M., Amir Ghodrati, and Tinne Tuytelaars.
TPAMI 2017.
PDF (arXiv:1512.01848) | IEEEXplore | Project website | Code | Bibtex

Motion Blur Characterization and Compensation for Line Scan (1D) Cameras.

José Oramas M., Agusmian P. Ompusunggu, Tinne Tuytelaars and Abdellatif B. Temsamani
SPIE Optical Metrology 2017 (oral).
Link | Bibtex

2016

Recovering Hard-to-Find Object Instances by Sampling Context-based Object Proposals.

José Oramas M. and Tinne Tuytelaars.
CVIU 2016.
Link | PDF (arXiv:1511.01954) | Bibtex

Modeling Visual Compatibility through Hierarchical Mid-level Elements.

José Oramas M. and Tinne Tuytelaars.
ECCV 2016 VSM (ws).
Ext.Abstract | Poster
PDF (arXiv:1604.00036) | Bibtex

Do Motion Boundaries Improve Semantic Segmentation?.

Yu-Hui Huang, José Oramas M. and Tinne Tuytelaars and Luc Van Gool.
ECCV 2016 (ws), NIPS 2016 (ws)
PDF | Bibtex


Reasoning about Body-Parts Relations for Sign Language Recognition.

Marc Martínez-Camarena*, José Oramas M.*, Mario Montagud Climent, and Tinne Tuytelaars.
* Denotes equal contribution
Technical Report 2016.
PDF (arXiv:1607.06356) | Bibtex

2015

Modeling Video Evolution For Action Recognition.

Basura Fernando, Efstratios Gavves, José Oramas M., Amir Ghodrati, and Tinne Tuytelaars.
CVPR 2015 (oral).
PDF | Ext.Abstract | Project website | Code | Bibtex

Towards Sign Language Recognition based on Body Parts Relations.

Marc Martínez-Camarena, José Oramas M., and Tinne Tuytelaars.
ICIP 2015 (oral).
PDF | Slides | Bibtex


Context-based reasoning for object detection and object pose estimation. (Ph.D. thesis).

José Oramas M.
Department of Electrical Engineering, KU Leuven.
April 2015.
PDF | Ext.Abstract | Slides | Bibtex

2014

Scene-driven Cues for Viewpoint Classification of Elongated Object Classes.

José Oramas M. and Tinne Tuytelaars.
BMVC 2014.
PDF | Supp.Material | Ext.Abstract | Poster | Bibtex


Towards Cautious Collective Inference for Object Verification.

José Oramas M., Luc De Raedt and Tinne Tuytelaars.
WACV 2014.
PDF | Slides | Poster | Bibtex


There are Plenty of Places like Home: Using Hierarchies and Relational Representations for Distance-based Image Understanding.

Laura Antanas, Martijn van Otterlo, José Oramas M., Tinne Tuytelaars and Luc De Raedt.
Neurocomputing 2014.
Link | Bibtex

2013

Allocentric Pose Estimation.

José Oramas M., Luc De Raedt and Tinne Tuytelaars.
ICCV 2013.
PDF | Poster | Bibtex


Rule-based Hand Posture Recognition Using Qualitative Finger Configurations Acquired with the Kinect

Lieven Billiet, José Oramas M., McElory Hoffmann, Wannes Meert and Laura Antanas,
ICPRAM 2013
PDF | Dataset | Demo-1 | Demo-2 | Bibtex

2012

A Relational Distance-based Framework for Hierarchical Image Understanding.

Laura Antanas, Martijn van Otterlo, José Oramas M. , Tinne Tuytelaars, Luc De Raedt.
ICPRAM 2012. (oral)(Best Paper Award)
PDF | Bibtex

2011

Potential benefits in the learning process of Ecuadorian Sign Language using a Sign Recognition System

José Oramas M. , Alejandro Moreno, Katherine Chiluiza.
eMinds 2011.
PDF | Bibtex

2010

Not far away from home: A relational distance-based approach to understand images of houses.

Laura Antanas, Martijn van Otterlo, José Oramas M., Tinne Tuytelaars, Luc De Raedt.
ILP 2010.
PDF | Bibtex

Projects

I have collaborated in the following projects:
- Tomorrow's Scalable and PersOnalised advertising Technology, Today (concluded)
- Lecture+ (concluded)
- ICO LoCoVision (concluded)
- IWT PARIS (concluded)
- ERC COGNIMUND (concluded)

Service

Reviewer for several scientific venues including TMM, CVIU, TIP, JMLC, JAIR, NIPS, ICML, CVPR, ICCV, ECCV, ICLR, AAAI, ICRA.
Co-organizer of the tutorial "Deep convolutional neural networks as a tool for vision science" at ECVP 2019
Session Chair at ICIP 2015.
Co-organizer of the Latin American Free Software Installation Festival (FLISOL) 2007-2009 at Guayaquil-Ecuador.
Co-founder of ESPOL’s Free Software community (KOKOA-ESPOL).

Technical Talks

07/11/2019 : "Interpreting and Explaining Deep Models Visually". BNAIC/BENELEARN'19, Brussels, Belgium.
25/09/2019 : "Towards Object Shape Translation Through Unsupervised Generative Deep Models". ICIP'19, Taipei, Taiwan.
25/08/2019 : "Model Interpretation and Explanation for Deep Neural Networks.". ECVP'19, Leuven, Belgium.
14/05/2019 : "Visual Explanation by Interpretation: Improving Visual Feedback Capabilities of Deep Neural Networks". TTIC, Chicago, US.
01/04/2019 : "Visual Explanation by Interpretation: Improving Visual Feedback Capabilities of Deep Neural Networks". PSI lab, KU Leuven. Leuven, Belgium.
31/05/2018 : "End-to-end modelling of visual data for autonomous navigation". LICT Workshop on autonomous Systems. Leuven, Belgium.
11/07/2016 : "Motion Blur Compensation for Line Scan Cameras". LoCoVision Hands-On Workshop, Flanders Make, Leuven, Belgium.
28/09/2015 : "Towards Sign Language Recognition based on Body Parts Relations". ICIP'15, Quebec City, Canada.
29/04/2015 : "Context-based Reasoning for Object Detection and Object Pose Estimation". Arenberg Castle, KU Leuven. Leuven, Belgium.
14/08/2014 : "Scene-driven Cues for Object Viewpoint Classification". PSI lab, KU Leuven. Leuven, Belgium.
19/05/2014 : "Towards Cautious Collective Inference for Object Verification". PSI lab, KU Leuven. Leuven, Belgium.
25/04/2014 : "Reasoning about Object Relations for Object Pose Classification". NCCV'14. Ermelo, Netherlands.
24/03/2014 : "Towards Cautious Collective Inference for Object Verification". WACV'14. Steamboat Springs, US.
26/11/2013 : "Allocentric Pose Estimation". PSI lab, KU Leuven. Leuven, Belgium.
04/06/2013 : "Cautious Collective Inference for Object Detection". CS. Dept., KU Leuven. Leuven, Belgium.
25/10/2010 : "Answering Complex Questions in Natural Language using Probabilistic Logic Programming and the Web". BNAIC'10. Luxembourg.
19/10/2009 : "Sign Language Recognition Technology for the Learning of Hearing-Impaired People". SIGHI.be'09. Belgium.