Ricardo Rossiter Barioni

Ricardo Barioni

Address
Recife/PE - Brazil

Contact
Email: rrbarioni@gmail.com
Phone: +55 81 98558-2677

 CV (Editor Tools GITEnglish and Portuguese)

Summary

Skilled Machine Learning Engineer with expertise in deep learning, audio/speech processing, NLP, computer vision, and academic research. Involved in projects across various fields such as human pose estimation, face recognition, and audio source classification. Interested in pursuing innovative solutions and continuous learning in the field of AI.


Experience

Machine Learning Engineer @ SiDi (Jan. 2021 - Current)

  • Involved in the development of NLP, audio and speech processing-based solutions.
  • Researched and published papers for the task of acoustic source classification.

Academic Researcher @ Voxar Labs (Aug. 2016 - Aug. 2020)

  • Carried out research in various fields and published academic papers.
  • Established partnerships with companies by providing research as a product.
  • Created a framework for quickly and easily creating human pose estimation datasets.
  • Researched state-of-the-art solutions for face recognition in images.
  • Led research on 3D object reconstruction solutions from RGB images.
  • Developed a tool for visualizing bat tracking data from thermal images.
  • Investigated the feasibility of augmented reality-based solutions for physiotherapy rehabilitation.

Education

M.Sc. in Computer Science (Aug. 2018 - Jul. 2020)

B.Sc. in Computer Science (Apr. 2014 - Jul. 2018)


Interests


Skills

Languages

  • Portuguese (native), English (fluent)

Software

  • Python, TensorFlow, PyTorch, Keras, OpenCV, NumPy, Matplotlib, Pandas, scikit-learn, Bash, SQLite, Git, Docker, LaTeX

Projects

HuTrain paper HuTrain video

HuTrain

path

This project is a framework for creating human pose estimation datasets quickly and easily. By using Python and libraries such as PyTorch and OpenCV, HuTrain comprises steps such as automatic camera calibration, refined human pose estimation and known dataset formats conversion.

Dog Breed Recognition Git

Dog Breed Recognition

path

This project is an algorithm for recognizing dog breeds from RGB images. By using Python and the PyTorch open-source machine learning framework, it applies convolutional neural network techniques for the classification of dog breeds and supports the enrolling of new dog breeds dynamically.

Credit Risk Analysis Git

Credit Risk Analysis

path

A project for the evaluation of the non-payment risk of bank clients. This credit risk analysis was implemented using Python and libraries such as Pandas, scikit-learn and Seaborn.

BalletVR paper BalletVR video

BalletVR

path

This system is a virtual reality application for guiding ballet dancers through learning and practicing basic ballet arm positions. By using a Microsoft Kinect for tracking the dancer's performed poses, the system compares them with basic arm positions, proposed by École Française, and allows the dancer to practice autonomously.

WRITEME WRITEME Git

WRITEME

path

This system consists of a web interface where developers can obtain recommendations of sections, based on research and the most popular open-source repositories, for the READMEs they are writing.

SongVerse paper SongVerse paper

SongVerse

path

This project is a Digital Music Instrument (DMI) that allows the user to create music in a virtual reality scenario where, by using wand controllers, the user interacts with an environment that resembles the outer space.

Onboarding Visualization Onboarding Visualization Git

Onboarding Visualization

path

This tool was built with the purpose of helping open-source maintainers to measure the effectiveness of their onboarding process, and give helpful tips on how to improve it.

Musical Invaders Musical Invaders Git

Musical Invaders

path

Based on the original 1978 arcade shooting game called Space Invaders, it is a web game where the player controls a spaceship, whose objective is to prevent aliens to reach earth by shooting musical notes. Not only fun, but Musical Invaders also encourages players to be creative by improvising new melodies while playing.

BatVis BatVis video BatVis Git

BatVis

path

This project is a web application for visualizing bats tracking data obtained from thermal images in caves. This application is able to provide insights, such as changes in bats populations and flight behavior, in a more intuitive fashion, which can be used to the biomonitoring of population tendencies, habitat use and the effects of climate change.

ARkanoidAR paper ARkanoidAR video

ARkanoidAR

path

This project is an augmented reality system that guides physiotherapy patients through the rehabilitation process of biomechanical movements at the sagittal plane. The system uses Microsoft Kinect for tracking the user's poses and instructs the user which movements must be performed by providing a series of visual and auditory feedback.


Publications

Zucatelli, Guilherme, et al. Improving Non-Stationary Acoustic Source Classification with Metric Learning

Zucatelli, Guilherme, et al. Non-Stationarity Objective Assessment for Acoustic Source Classification

Zucatelli, Guilherme, et al. A Metric Learning Based Solution for Non-Stationary Acoustic Source Classification

Barioni, Ricardo R., et al. HuTrain: a Framework for Fast Creation of Real Human Pose Datasets

Costa, Willams, et al. Songverse: a digital musical instrument based on Virtual Reality

Cavalcanti, Virgínia C., et al. Usability and effects of text, image and audio feedback on exercise correction during augmented reality based motor rehabilitation

Barioni, Ricardo R., et al. BalletVR: a Virtual Reality System for Ballet Arm Positions Training

  • Full paper at 2019 21st Symposium on Virtual and Augmented Reality (SVR)

Costa, Willams, et al. Songverse: a music-loop authoring tool based on Virtual Reality

  • Full paper at 2019 21st Symposium on Virtual and Augmented Reality (SVR)

Barioni, Ricardo R., et al. Human Pose Tracking from RGB Inputs

  • Full paper at 2018 20th Symposium on Virtual and Augmented Reality (SVR)

Santana, Maria I., et al. ARkanoidAR 2.0: Otimizações em uma solução de realidade aumentada com base em testes de usabilidade

  • Poster at 2018 26th Congresso Brasileiro de Engenharia Biomédica (CBEB)

Barioni, Ricardo R., et al. ARkanoidAR: an Augmented Reality System to Guide Biomechanical Movements at Sagittal Plane

  • Full paper at 2017 19th Symposium on Virtual and Augmented Reality (SVR)

Certificates

Deploying Machine Learning Models in Production

Machine Learning Modeling Pipelines in Production

Probability & Statistics for Machine Learning & Data Science

Introduction to Embedded Machine Learning

Machine Learning Data Lifecycle in Production

Types of Conflict

Mathematics for Machine Learning: Linear Algebra

Conflict Resolution Skills

Communication in the 21st Century Workplace

Effective Problem-Solving and Decision-Making

Work Smarter, Not Harder: Time Management for Personal & Professional Productivity

Digital Signal Processing 1: Basic Concepts and Algorithms

Device-based Models with TensorFlow Lite

Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization

Introduction to Machine Learning in Production

Sequence Models


Leaderships and Awards