MY SITEACADEMIC & PERSONAL INFO
.01

ABOUT

PERSONAL DETAILS
Richmond, London
contact@patrycia.co.uk
Hello! I’m a Software Engineer by profession and a Programmer by heart, with a passion for cameras and cute robots. Welcome to my World (Wide Web) corner!

SHORT BIOGRAPHY

ABOUT ME

I’m a Software Engineer with interest in Computer Vision, Robotics and Machine Learning. I received a Master of Science Degree in Computer Vision and Robotics in 2012 from the University of Burgundy (UB).

I’ve accumulated twenty years of experience working in Software Development, in which I undertook many different projects, from very simple desktop applications to highly distributed and data intensive back-end systems.

I’m settled in London, UK, and my major goal in life is to become a Professor and conduct research in Robotics and Machine Learning for applications in the healthcare industry.

HOBBIES

INTERESTS

I play the piano as a hobby. I started when I was seven years old. However, by my twenties I decided to concentrate in my career only. It was more of the same in those times: no time or money to support my hobby.

Recently, I bought a digital piano and I’m back to my piano lessons. It's never too late to learn how to play an instrument, specially if you love music like I do.

Playing piano

FACTS

FOUR FACTS ABOUT ME

Cup of Coffee

There is always a cup of coffee in my desk when I’m working. People prefer tea in England. Maybe, I should write a program to guess nationalities based on the “cup of something” people have in their desks.

Musician or Geek

I thought about being a professional musician once. When I was eighteen years old I was admitted in a renowned music University in my city. At the same time, I was offered a place in another University to study computer science. Maybe, the best way to summarise my life is: She loves a keyboard.

Citizen of the world

When I was young, I dreamed about going to different places or living abroad for a while. My dream became reality and I’ve already lived in four different countries (not counting my own). I wonder what else life will bring to my future.

Gadgets

I’m passionate about cameras and robots and I have a total of six cameras e two robots which I use in my research projects. Ok, you can say I’m a gadget person! Who isn’t?

.02

RESUME

EDUCATION
  • 2012
    2010
    Dijon, France

    Master of Science in Computer Science and Electronic, Specialist in Image, Informatics and Engineering

    University of Burgundy (UB)

    International master's program, part of the ERASMUS MUNDUS MSC/VIBOT2 program in Computer Vision and Robotics. The course was taught in English and comprised by 2 years of study and the presentation of a final research project/thesis entitled: “Gestalt-inspired feature extraction for object category recognition”. Supervisors: Dr. Fabrice Meriaudeau and Dr. Alamin Mansouri.
  • 2010
    2009
    Brasilia, Brazil

    Postgraduate Diploma in Clinical Engineering

    University of Brasilia (UNB)

    A specialisation program on the use of technology in healthcare institutions, with a total of 12 months of study and a final project entitled: “A study for RIS, PACS and digital mammography equipment integration”. Supervisor: Dr. Lourdes Mattos Brasil.
  • 2007
    2006
    Brasilia, Brazil

    Postgraduate Diploma in Software Engineering

    Catholic University of Brasilia (UCB)

    A specialisation program with a total of 12 months of study and a final project entitled “WRIM-WS: A model based on EBXML to provide the register and repository of Web Services”. Supervisor: MSc. Prof. Claudia Melo.
  • 2006
    Brasilia, Brazil

    Technician Degree in Data Processing

    Catholic University of Brasilia (UCB)

    Higher education degree in Data Processing/Computer Science.
AWARDS AND RECOGNITION
  • June 2006
    Brasilia, Brazil

    Best Graduate Student Paper

    Catholic University – XIV IT Workshop

    REQBOOK – A Practical Guide for Requirements Engineering
  • August 2005
    Brasilia, Brazil

    Best National e-Gov Software Project on the Government-to-Business Category (G2N)

    e-Gov 2005 Conference held by the Brazilian Association of State IT and ICT Entities (ABEP)

    Project CNES (National Record of Union Entities) was developed for the Ministry of Labor in 2004/2005 and has been updated frequently due to changes in law. However, its conceptual idea and technical architecture are the same and provides information to more than 15.000 entities in Brazil.
ACADEMIC JOBS
  • CURRENT
    2015
    Brasilia, Brazil

    RESEARCH ASSISTANT (COMPUTER VISION)

    Research Laboratory on Informatics in Healthcare (LIS), University of Brasilia (UnB)

    Conducts research on the development of a software to assist the treatment of children diagnosed with Autism Spectrum Disorder (ASD). Her responsibility is to design and program the NAO Robot’s teleoperation module in Unity 3D and C# by using motion sensors and Kinect camera. Participates in the study groups of “Robotics and Nanorobots” and “Informatics Applied to Healthcare and Medicine”.
  • 2015
    2012
    London, UK

    RESEARCH ASSISTANT (SYSTEM INTEGRATION & 3D RECONSTRUCTION)

    Queen Mary, University of London

    Worked on the development of REVERIE software, a FP7 European project, for “Real-Time and Virtual Engagement in Realistic Immersive Environments”. Contributed on the development of a multi-camera calibration tool and 3D reconstruction of scenes and human-like avatars using C++, OpenGL, OpenCV, QT and RGBD Microsoft Kinect Sensors (version 1 and 2).
  • 2010
    2009
    Brasilia, Brazil

    RESEARCH ASSISTANT

    Research Laboratory on Informatics in Healthcare (LIS), University of Brasilia (UnB)

    Worked on the development of the “Anatomical 3D Atlas Applied to the Female Breast” using C++ and OpenGL, as well as, the design of an ontological model to describe the female breast anatomy structure in OWL, Protégé and OWL Java API. Developed a face recognition tool applying Active Appearance Model (AAM) in MATLAB, which resulted on a book chapter published by IGI Global in 2013.
OTHER JOBS
  • 2010
    1998
    Brasilia, Brazil

    ENTREPRENEUR/SENIOR SOFTWARE ENGINEER

    Visual Code Inc.

    Opened a start-up company to provided IT services (software development and systems integration) for industries in Brazil and agencies of the Brazilian Federal Government. Provided training/tutoring on those subjects and on the use of specialised tools for Service-Oriented Architecture (SOA) and Data Mining, such as: Oracle Sunopsis ODI Tool and Oracle Middleware Integrator.
  • 1998
    1993
    Brasilia, Brazil

    SOFTWARE ENGINEER/DEVELOPER

    Several companies/institutions in Brazil

    Worked as a developer for several private companies in Brasilia, Brazil. Applied programming skills in Java, Active Server Pages (ASP) and Visual Basic, building desktop and web applications. Worked as a system analyst to document software requirements and help on the design of system architectures.
.03

TEACHING

TEACHING HISTORY
  • 2010
    2009
    Brasilia, Brazil

    JAVA/C++ TEACHER

    Paulista University (UNIP)

    Worked as an external instructor, teaching undergraduate students the basics of programming in Java and C++.
  • 2010
    2009
    Brasilia, Brazil

    JAVA TEACHER (OBJECT-ORIENTED PROGRAMMING & WEB DEVELOPMENT)

    University UNIEURO

    Worked as an external instructor, teaching undergraduate students advanced programming skills in Java, such as: Object-Oriented Programming, Web Development with JavaServer Pages (JSP) and Software Architecture for the Web.
  • 2010
    2000
    Brasilia, Brazil

    TEACHER (SOA ARCHITECTURE & SYSTEM INTEGRATION)

    Private Companies and Government Agencies

    Worked as an instructor on SOA and system integration tools, such as, BEA/Oracle BPM Suite, Sunopsis/Oracle ODI, Oracle Integrator/Fusion Middleware and BPEL Process Manager.
  • 2008
    2007

    JAVA TEACHER

    National Service for Industrial Training (SENAI)

    Delivered training courses in Java as part of the Professional Qualification Program for members of the industrial sector in Brazil.
.04

PUBLICATIONS

PUBLICATIONS LIST
10 Dec 2014

ACCURATE STEREO 3D POINT CLOUD GENERATION SUITABLE FOR MULTI-VIEW STEREO RECONSTRUCTION

IEEE Visual Communications and Image Processing (VCIP 2014)

This paper proposes a novel methodology for generating 3D point clouds of good accuracy from stereo pairs RGB images.

Conferences Selected G.A. Kordelas, P. Daras, P. Klavdianos, E. Izquierdo and Q. Zhang

ACCURATE STEREO 3D POINT CLOUD GENERATION SUITABLE FOR MULTI-VIEW STEREO RECONSTRUCTION

G.A. Kordelas, P. Daras, P. Klavdianos, E. Izquierdo and Q. Zhang Conferences Selected

This paper proposes a novel methodology for generating 3D point clouds of good accuracy from stereo pairs RGB images. Initially, the methodology defines some conditions for the proper selection of image pairs. Then, the selected stereo images are used to estimate dense correspondences using the Daisy descriptor. An efficient two-phase strategy to remove outliers is then introduced. Finally, the 3D point cloud is refined by combining sub-pixel accuracy correspondences estimation and the moving least squares algorithm. The proposed methodology can be exploited by multi-view stereo algorithms due to its good accuracy and its fast computation.

05 Jul 2013

A CONCISE SURVEY FOR 3D RECONSTRUCTION OF BUILDING FACADES

14th International Workshop on Image Analysis for Multimedia Interactive Services (WIAMIS 2013)

This paper presents a review for 3D reconstruction of building facades.

Conferences P. Klavdianos, Q. Zhang and E. Izquierdo

A CONCISE SURVEY FOR 3D RECONSTRUCTION OF BUILDING FACADES

P. Klavdianos, Q. Zhang and E. Izquierdo Conferences

3D facade modeling consists of representing architectural elements of a building in detail so that both geometry and appearance are generated in a photo-realistic 3D scene. In this article, we review four main techniques used to solve this problem: photogrammetry, sparse and dense reconstruction based on SfM (Structure from Motion) and procedural modeling. We provide a comparison of several methods in these categories by considering their main advantages and limitations.

18 Sep 2013

GESTALT-INSPIRED FEATURES EXTRACTION FOR OBJECT CATEGORY RECOGNITION

20th IEEE International Conference on Image Processing (ICIP 2013)

This paper proposes a methodology inspired by Gestalt laws to extract and combine features for object category recognition problem.

Conferences P. Klavdianos, A. Mansouri and F. Meriaudeau

GESTALT-INSPIRED FEATURES EXTRACTION FOR OBJECT CATEGORY RECOGNITION

P. Klavdianos, A. Mansouri and F. Meriaudeau Conferences

We propose a methodology inspired by Gestalt laws to extract and combine features and we test it on the object category recognition problem. Gestalt is a psycho-visual theory of Perceptual Organization that aims to explain how visual information is organized by our brain. We interpreted its laws of homogeneity and continuation in link with shape and color to devise new features beyond the classical proximity and similarity laws. The shape of the object is analyzed based on its skeleton (good continuation) and as a measure of homogeneity, we propose self-similarity enclosed within shape computed at super-pixel level. Furthermore, we propose a framework to combine these features in different ways and we test it on Caltech 101 database. The results are good and show that such an approach improves objectively the efficiency in the task of object category recognition.

29 Apr 2013

MODELING THE ARCHITECTURE OF AN INTELLIGENT TUTORING SYSTEM APPLIED TO BREAST CANCER

International Conference, Pan American Health Care Exchanges (PAHCE 2013)

This paper presents an Intelligent Tutoring System (ITS) applied to the teaching of anatomy of the female breast, including some types of cancer related to this organ.

Conferences H.P. Maffon, L.M. Brasil, J.S. Melo, P.B. Klavdianos, G. Curilem and J.M. Lamas

MODELING THE ARCHITECTURE OF AN INTELLIGENT TUTORING SYSTEM APPLIED TO BREAST CANCER

H.P. Maffon, L.M. Brasil, J.S. Melo, P.B. Klavdianos, G. Curilem and J.M. Lamas Conferences

This paper presents an Intelligent Tutoring System (ITS) applied to the teaching of anatomy of the female breast, including some types of cancer related to this organ. This ITS is composed of four modules: Student, an Expert System containing a questionnaire for the diagnosis of the learner’s profile; Tutor, an Artificial Neural Network Interactive Activation and Competition, for the application of teaching techniques; Domain, some ontologies containing content and related media; and Interface, developed as an Adaptive Hypermedia System. The objective of this work is to lift requirements for the integration of various modeling types that use Artificial Intelligence techniques in the same ITS, even enabling the use of this system in a Medical Simulation Environment. The validation process of this ITS is in progress because the class period has not yet started at three universities: the University of Brasilia (Federal) and the Catholic University of Brasília (Private), both located in Brasilia, Brazil, and the Universidad de La Frontera (Federal) located in Temuco, Chile.

09 Dec 2013

FACE RECOGNITION WITH ACTIVE APPEARANCE MODEL (AAM)

IGI Global: International Publisher of books, journals and cases related to technology and science

Chapter of the book «Robotic Vision: Technologies for Machine Learning and Vision Applications» that describes the use of Active Appearance Model (AAM) to solve the problem of face recognition.

Book Chapters P.B.L. Klavdianos, L.M. Brasil and J.S.S Melo

FACE RECOGNITION WITH ACTIVE APPEARANCE MODEL (AAM)

P.B.L. Klavdianos, L.M. Brasil and J.S.S Melo Book Chapters

Recognition of human faces has been a fascinating subject in research field for many years. It is considered a multidisciplinary field because includes understanding different domains such as psychology, neuroscience, computer vision, artificial intelligence, mathematics and many others. Human face perception is intriguing and draws our attention because we accomplish the task so well that we hope one day witnessing a machine performing the same task in a similar or better way. This chapter aims to provide a systematic and practical approach regarding to one of the most current techniques applied on face recognition, known as AAM (Active Appearance Model). AAM method will be addressed considering 2D face processing only.  This chapter doesn’t cover the entire theme, but offers to the reader the necessary tools to construct a consistent and productive pathway toward this involving subject.

21 Jul 2012

GESTALT-INSPIRED FEATURES EXTRACTION FOR OBJECT CATEGORY RECOGNITION

Laboratoire Electronique, Informatique et Image (LE2I), University of Burgundy

Master's thesis presented in 2012 that proposes a novel method, inspired by Gestalt laws, to extract and combine features for the object category recognition problem.

Theses P.B.L. Klavdianos

GESTALT-INSPIRED FEATURES EXTRACTION FOR OBJECT CATEGORY RECOGNITION

P.B.L. Klavdianos Theses

Our perception about the word is so deeply ingrained in our day life experiences that we never stop to ask ourselves how all the pieces of visual information are organized in large and small units in order to give birth to the things we so much appreciate in our world. This is the problem addressed by perceptual organization originated with the Gestalt psychologist in 1935. Nowadays it is also a problem of interest of Computer Vision practitioners which aim to building system which perceive and understand the world. In this work we propose a new method for extracting features from images, inspired in the Gestalt laws and for the purpose of object categorization. Our main interest is to provide an organization of visual information in higher semantical levels and test its eciency in the categorization problem. Most of the work in this area consider feature extraction on the pixel-level and applying the Gestalt laws of proximity and similarity only. Our approach extends this idea by considering other Gestalt principles and adding new features in order to capture not only semantic information about the shape of the object, but also regarding its appearance.

04 Feb 2012

ONTO-MAMA: AN UNIFIED ONTOLOGY AND 3D GRAPHIC MODEL OF THE FEMALE BREAST ANATOMY

International Conference on Health Informatics (BIOSTEC 2012)

This paper presents the creation and integration of an ontology model and a 3D graphic model of the female breast anatomy to be used in a virtual reality environment that assist learners in the practice of the core needle biopsy.

Conferences P.B.L. Klavdianos, M. Parente, L.M. Brasil and J.M. Lamas

ONTO-MAMA: AN UNIFIED ONTOLOGY AND 3D GRAPHIC MODEL OF THE FEMALE BREAST ANATOMY

P.B.L. Klavdianos, M. Parente, L.M. Brasil and J.M. Lamas Conferences

The science of ontology has been widely used in knowledge management, either to define and organize concepts for the purpose of future preservation or to provide more efficient information processing by our computers. From this, the use of ontology in medical field has become increasingly consolidated, either to simply describe the correct meaning of technical terms or to completely depict the anatomical structure of the human body or medical procedures. In addition, the use of ontology has been associated to artificial intelligence and virtual reality in order to provide simulation of medical environments with the aim of better understand the complexities of the human anatomy and the medical procedures. Accordingly, this article presents the elaboration of an ontology and a 3D graphic model of the female breast anatomy to be used in a virtual reality environment containing an intelligent tutor system which will eventually be able to assist learners in the practice of the core needle biopsy. This article reports our experience so as to share information about the process used, the artifacts generated and the systematic involved in the structuring of such unified model.

28 Mar 2011

ONTO-MAMA: AN ONTOLOGY OF THE FEMALE BREAST ANATOMY APPLICABLE TO A VIRTUAL LEARNING ENVIRONMENT (in Portuguese)

International Conference, Pan American Health Care Exchanges (PAHCE 2011)

This paper explores the use of an ontology model to describe the anatomy of the female breast structure. As a target application, this project proposes to build a virtual reality environment that implements an intelligent tutor system to assist learners in the practice of core needle biopsy surgery.

Conferences P.B.L. Klavdianos, E.K.F. Souza, L.M. Brasil and J.M. Lamas

ONTO-MAMA: AN ONTOLOGY OF THE FEMALE BREAST ANATOMY APPLICABLE TO A VIRTUAL LEARNING ENVIRONMENT (in Portuguese)

P.B.L. Klavdianos, E.K.F. Souza, L.M. Brasil and J.M. Lamas Conferences

The ontology has been widely used in knowledge management, either with a view to defining and organizing concepts in order to preserve it for future generations or to allow for a more efficient processing of information by computers. This paper presents the elaboration process of an ontology to describe the anatomy of the female breast to be used in a virtual reality environment containing an intelligent tutor system that will eventually be able to assist learners in the practice of core needle biopsy.

05 Dec 2015

INTEROPERABILITY GUIDE OF THE BRAZILIAN GOVERNMENT: TECHNICAL REPORT FOR DEVELOPERS (in Portuguese)

Ministry of Planning, Budget and Management (e-Ping)

This technical manual presents good practices on using technology for e-Government interoperability under the point of view of developers and software engineers.

Technical Report P.B.L. Klavdianos et al.

INTEROPERABILITY GUIDE OF THE BRAZILIAN GOVERNMENT: TECHNICAL REPORT FOR DEVELOPERS (in Portuguese)

P.B.L. Klavdianos et al. Technical Report

Technical report about e-Government interoperability that presents technical requirements and suggests use of technologies and standards with the objective of promoting system integration and better quality on the development of e-Gov solutions.

04 Dec 2015

INTEROPERABILITY GUIDE OF THE BRAZILIAN GOVERNMENT: TECHNICAL REPORT FOR MANAGERS (in Portuguese)

Ministry of Planning, Budget and Management (e-Ping)

This technical manual presents good practices on using technology for e-Government interoperability under the point of view of IT managers.

Technical Report P.B.L. Klavdianos et al.

INTEROPERABILITY GUIDE OF THE BRAZILIAN GOVERNMENT: TECHNICAL REPORT FOR MANAGERS (in Portuguese)

P.B.L. Klavdianos et al. Technical Report

Technical report about e-Government interoperability under the point of view of an IT Manager who works on the development of solutions for the Brazilian Government.

28 Mar 2011

A PROCESS FOR INTEGRATION OF RIS, PACS AND DIGITAL MAMMOGRAPHY EQUIPMENT (in Portuguese)

XXII Brazilian Congress of Biomedical Engineering (CBEB 2010)

This paper describes a process for integrating technologies used in clinics and hospitals, such as: RIS (Radiology Information Systems), PACS (Picture Archiving and Communications Systems) and digital mammography equipments.

Conferences P.B.L. Klavdianos, J.M. Lamas and L.M. Brasil

A PROCESS FOR INTEGRATION OF RIS, PACS AND DIGITAL MAMMOGRAPHY EQUIPMENT (in Portuguese)

P.B.L. Klavdianos, J.M. Lamas and L.M. Brasil Conferences

The transition from Mammography based on films to Digital Mammography or FFDM (Full-field Digital Mammography) is not a simple task, involving solely the replacement of a few electronic components and the addition of new computers and software. This paper describes a process for integrating RIS (Radiology Information Systems), PACS (Picture Archiving and Communications Systems) and digital mammography equipments in order to help clinical engineers and IT professionals in planning, preparation and implementation of a FFDM environment in conformity with the data communication standards used in the medicine field.

30 Oct 2010

A STUDY FOR INTEGRATION OF RIS, PACS AND DIGITAL MAMMOGRAPHY EQUIPMENT

University of Brasilia (UnB)

A dissertation submitted in partial satisfaction of the requirements for the PGDip in Clinical Engineering. Provides a study for the integration between RIS (Radiology Information Systems), PACS (Picturing Archiving and Communications Systems) and digital mammography equipments.

Theses P.B.L. Klavdianos

A STUDY FOR INTEGRATION OF RIS, PACS AND DIGITAL MAMMOGRAPHY EQUIPMENT

P.B.L. Klavdianos Theses

This work offers a study involving the integration between RIS (Radiology Information Systems), PACS (Picturing Archiving and Communications Systems) and digital mammography equipments aiming at helping clinical engineers and IT professionals in the planning, preparation and implementation of an FFDM (Full-field Digital Mammography) environment in conformity with the data communication standards used in the medicine field nowadays and, more specifically, in the digital mammography practice.

01 Dec 2008

REQBOOK: A PRACTICAL GUIDE FOR REQUIREMENTS ENGINEERING (in Portuguese)

XIV Workshop of Information Technology at Catholic University of Brasília

This work provides a practical guide for requirement engineering.

Conferences P.B.L. Klavdianos, L. Neves and C. Melo

REQBOOK: A PRACTICAL GUIDE FOR REQUIREMENTS ENGINEERING (in Portuguese)

P.B.L. Klavdianos, L. Neves and C. Melo Conferences

This work provides a practical guide for requirement engineering which comprises the following processes: preparation for requirement management, problem identification and analysis, understanding users need, defining the problem, scope management, requirements refinement and managing requirements change.

11 Jul 2007

WRIM-WS: A MODEL BASED ON EBXML TO PROVIDE THE REGISTER AND REPOSITORY OF WEB SERVICES (in Portuguese)

Catholic University of Brasilia (UCB)

A dissertation submitted in partial satisfaction of the requirements for the PGDip in Software Engineering.

Theses P.B.L. Klavdianos

WRIM-WS: A MODEL BASED ON EBXML TO PROVIDE THE REGISTER AND REPOSITORY OF WEB SERVICES (in Portuguese)

P.B.L. Klavdianos Theses

The rise in the use of Web Services imposes the need to create new mechanisms to describe and store information about the services offered and their providers. Such information should be part of a centralised or federated catalogue, whose main goal would be to document the offered services on a patterned fashion, making searches easier and enabling their proper management (i.e. policy management, service level agreements, pricing and charge-back models). This dissertation provides a study about Services Oriented Architecture (SOA) and the use of Web Services technology. A meta-data model denominated XRIM-WS was created based on the ebXML-RIM standard, as to provide a detailed and conceptual specification to develop a register/repository of Web Services.

01 Jul 2007

THE REFERENTIAL ARCHITECTURE AND THE E-PING: THE BRAZILIAN GOVERNMENT INITIATIVE ON SOA

E-Government Interoperability Campus (eGovInterop 2007), France, Paris

This paper depicts the Referential Architecture (AR) - the Brazilian Government initiative on SOA.

Conferences E. Franzosi, P.B.L. Klavdianos, C. Mesquita and E. Favero

THE REFERENTIAL ARCHITECTURE AND THE E-PING: THE BRAZILIAN GOVERNMENT INITIATIVE ON SOA

E. Franzosi, P.B.L. Klavdianos, C. Mesquita and E. Favero Conferences

This paper depicts the Referential Architecture (AR) – the Brazilian Government initiative on SOA. AR uses the Electronic Government Interoperability Standards (e-PING) as a guide to implement technological IT standards. Within a logical and physical view, AR includes a QoS3 model to be applied on real service-based governmental projects. It also details a case study of AR to illustrate the operation and applicability of such model.

.05

RESEARCH

MEET MY TEAM

Currently, I'm working for the Laboratory on Informatics in Healthcare (LIS) at the University of Brasilia (UnB), Brazil. I'm conducting a research on the application of assistive tools (AT) for the treatment of teenagers (age of 10 to 18) diagnosed with mild forms of Autism Spectrum Disorder (ASD). This is my team:

Prof. Dr. Lourdes Mattos Brasil

Coordinator of the Project and Director of LIS

Lídia Isabel Barros dos Santos

Therapist

NAO

Our Robot

We are investigating the use of virtual reality and robotics as intervention mechanisms to facilitate the work of a therapist in training social abilities of individuals with ASD. The project makes use of Unity3D, Vive HTC Virtual Reality Kit, Perception Neuron motion capture technology and NAO Robots.

I'm working with the Perception Neuron motion capture technology to train the robot using an imitative approach. The main idea is to enable the therapist to construct several scenarios by training a robot to be social and to respond accordingly in a more realistic and autonomous way during a therapy session. As a result, the robot can operate as the child's companion and friend, which we believe will make the treatment more natural and pleasant.

The diagram below illustrates the modules I idealized for this work, going from the robot training to the execution in a real-world scenario. We expect to run our first trial in the end of next year with the support of the Federal District State Health Bureau in Brasilia, Brazil.

System Modules
.06

SKILLS

SOFTWARE DEVELOPMENT
Programming Skills > I'm very confident working with Java, C, C++ and C#. Recently, I ventured into Python. As some people say: "To learn a new language, first you need to fall in love with it". Python is an amazing language to learn. But, I think I fell in love with programming language, in general, long time ago. Can't wait to use my Python skills to build cool data science applications. 😉
LEVEL : ADVANCED EXPERIENCE : 20 YEARS
Java C, C++, C# Visual Basic Python
Web Development > Over the last 10 years I've been developing software for the web. I'm very comfortable designing solutions for this environment, specially using Java technology on distributed architecture based on Service Oriented Architecture (SOA) and Business Objects running on the server side or in the Cloud.
LEVEL : ADVANCED EXPERIENCE : 15 YEARS
Java, JSP, JavaBeans ASP SOA, Web Services HTML5, CSS, JavaScript
INTEGRATED DEVELOPMENT ENVIRONMENTS (IDE) > I love using IDEs as part of my work and I've tried many already, including proprietary and opensource ones. When I work with C, C++ and C#, Visual Studio is my favourite IDE. For Java projects, I prefer to use the old and good Eclipse, although NetBeans is also a good choice.
LEVEL : ADVANCED EXPERIENCE : 20 Years
Visual Studio Eclipse NetBeans
COMPUTER VISION AND ROBOTICS
Computer Vision > I hold a masters degree in computer vision applied to the healthcare industry and since my graduation I've been working in research projects that combines computer vision and 3D graphics. I had the opportunity to use several frameworks and open standards in my work, but I'd like to highlight the followings: OpenGL, OpenCV, Point Cloud Library (PCL). Although I feel very confident in this field, there are many things I still have to learn. Machine Learning is one of those skills that I hope to master in few years.
LEVEL : INTERMEDIATE EXPERIENCE : 5
OpenGL OpenCV PCL
Robotics > During my masters I had introductory courses about robotics and since then I'm learning on my own. Currently, I'm working with the NAO Robot in which I program in C++ and Python, using its API. One day I'd be able to combine my programming skills, computer vision and robotics to build applications for the healthcare industry.
LEVEL : BEGINNER EXPERIENCE : Hobbyist and Curious
NAOqi APIs
.07

WORKS

MY PORTFOLIO
COMPUTER VISION

MULTI-CAMERA CALIBRATION

MULTI-CAMERA CALIBRATION

This work involved the development of a complete tool for camera calibration in a multi-sensor setup. The internal calibration of a single Kinect involves the estimation of (1) the RGB’s camera intrinsic parameters, (2) the depth camera’s intrinsic parameters and (3) the relative pose between them.

The OpenNI API uses the factory calibration, stored on board. For this application the data could be considered as sufficiently accurate. Since, however, the external calibration of the Kinects network is based on RGB data, in order to increase accuracy, it was estimated the internal RGB cameras’ parameters using Zhangs method, through homography obtained from a checkerboard.

As for the estimation of the extrinsic parameters, it was employed a global (all-to-all) external calibration method. The method requires establishing point correspondences across all camera images. For that purpose, it was used a calibration object with two LEDs of different colors (a red and a green one), positioned at a fixed distance. The employed approach can be shortly summarized in the following steps: (1) Establishment of point correspondences, (2) Pairwise calibration, and (3) Global optimization.

Estimations of the 3D LED points are initially produced by averaging the pairwise stereo-triangulation estimates. Then, the 3-D point estimates (structure term), as well as the extrinsic camera parameters (motion term) are globally refined using the Sparse Bundle Adjustment implementation. Finally, given the globally optimized 3D point structure, the scale factors are re fined.

COMPUTER VISIONVIRTUAL REALITY

REAL-TIME 3D RECONSTRUCTION OF AVATARS

REAL-TIME 3D RECONSTRUCTION OF AVATARS

In this project it was used a multi-sensor camera setup composed of maximum M=3 Kinect1 or Kinect2 sensors, interconnected by four different machines. The sensor kinect2 requires a dedicated USB3 port on each machine and it doesn’t allow the use of two or more sensors even if the machine has more than one USB3 port available. This is due to the fact that kinect2 needs to use a dedicated daemon program to communicate with the sensor. The daemon program is installed on windows 8.0 as part of the kinect studio sdk and runs as a service in the operational system. It is possible in this setup to use four dedicated machines for data capturing and one, more powerful, machine for the 3D reconstruction.

The frontal lateral devices are positioned at the vertices of an imaginable octahedron while the frontal and back cameras are located in the edges of the front and back face of the octahedron. This
arrangement guarantees an overlap of images captures from the frontal sensor and the lateral right and left ones. This overlap is a requirement in our experiment of 3D reconstruction using only
depth information. This setup suggests a diagonal of length approximately 1.5 meters, at a height of approximately 1.1 meters and pointing the devices to the center of the working volume, considering that the depth cameras’ FOV (fi eld of view) is approximately 45 and a mean human height of 1.7 meters. The setup introduces an “active” region of approximately 1.4 x 1.4 square meters, where the whole human body can be perfectly captured.

Two or more users using distinct systems can share the same scene and interact with each other through audio. The camera calibration is performed as an off-line process.

COMPUTER VISIONVIRTUAL REALITY

REAL-TIME BODY MOTION CAPTURE WITH KINECT2

REAL-TIME BODY MOTION CAPTURE WITH KINECT2

In this project, a kinect2 sensor is used to capture the entire motion of a human body. The movement is immediately transferred to a 3D mesh avatar which allows the user to play the game SIMON SAYS in real-time.

 

COMPUTER VISIONVIRTUAL REALITY

3D RECONSTRUCTION OF SCENES

3D RECONSTRUCTION OF SCENES

This work proposes a novel methodology for generating 3D point clouds of good accuracy from stereo pairs. Initially, the methodology defines some conditions for the proper selection of image pairs. Then, the selected stereo images are used to estimate dense correspondences using the Daisy descriptor. An efficient two-phase strategy to remove outliers is then introduced.

Finally, the 3D point cloud is refined by combining sub-pixel accuracy correspondences estimation and the moving least squares algorithm. The proposed methodology can be exploited by multiview stereo algorithms due to its good accuracy and its fast computation.

.08

CONTACT

Drop me a line

GET IN TOUCH

Use the contact form below if you have any questions, requests, suggestions, constructive criticisms or general comments. I promise to respond to your message as soon as possible.