Jose Luis Flores Campana

Jose Luis Flores Campana

Ph.D. in Computer Science

University of Campinas

Biography

Jose Luis Flores received his B.Sc. in Computer and Software Engineering from the University of San Antonio Abad de Cusco (UNSAAC), Peru, in 2016. As a bachelor’s student, Jose worked on a research paper related to the recognition and classification of hand gestures based on sign language using artisanal and deep learning techniques. After. Jose obtained his M.Sc in Computer Science from the State University of Campinas (Unicamp), Brazil, in 2020. As a master’s student, Jose was part of a team of researchers from SAMSUNG Brasil and UNICAMP. In this team he worked on two projects, “Multilingual text detection and recognition in images and videos” and “Generation of parallax motion effects”. In 2024, Jose received his Ph.D. from the State University of Campinas (Unicamp), Brazil. As a Ph.D. student, Jose worked on topics such as Image Inpainting and Image Synthesis, focusing his research on Deep Learning models such as Generative Adversarial Networks and Vision Transformer. His research focuses on Machine Learning, Deep Learning, and Image Processing, with specialization in Text Detection and Recognition in images and videos, Image Inpainting, and Image Synthesis. He currently works as a software engineer at Loggi.

Jose Luis Flores recebeu seu B.Sc. em Engenharia de Computação e Software pela Universidade de San Antonio Abad de Cusco (UNSAAC), Peru, em 2016. Como estudante de bacharelado, Jose trabalhou em um trabalho de pesquisa relacionado ao reconhecimento e classificação de gestos manuais com base na linguagem de sinais usando técnicas de aprendizamgem artesanais e profundas. Depois. Jose obteve seu M.Sc. em Ciência da Computação pela Universidade Estadual de Campinas (Unicamp), Brasil, em 2020. Como aluno de mestrado, Jose fez parte de uma equipe de pesquisadores da SAMSUNG Brasil e da UNICAMP. Nesta equipe trabalhou em dois projetos, “Detecção e reconhecimento de texto multilíngue em imagens e vídeos” e “Geração de efeitos de movimento paralaxe”. Em 2024, Jose recebeu o Ph.D. pela Universidade Estadual de Campinas (Unicamp), Brasil. Como estudante de Ph.D., Jose trabalhou em temas como Image Inpainting e Image Synthesis, focando sua pesquisa em modelos de Deep Learning tais como Generative Adversarial Networks e Vision Transformer. Sua pesquisa se concentra em Aprendizado de Máquina, Aprendizado Profundo e Processamento de Imagens, com especialização em Detecção e Reconhecimento de Texto em imagens e vídeos, Pintura de Imagens e Síntese de Imagens. Atualmente trabalha como engenheiro de software na Loggi.

Download my resumé.

Interests
  • Pattern Recognition
  • Computer Vision
  • Image Processing
  • Image Synthesis
  • Machine Learning
  • Deep Learning
Education
  • Ph.D. in Computer Science, 2024

    University of Campinas (IC/Unicamp)

  • M.Sc. in Computer Science, 2020

    University of Campinas (IC/Unicamp)

  • B.Sc. in Computer Engineering, 2017

    University San Antonio Abad of Cusco (UNSAAC)

Skills

Git

70%

Github

70%

Docker

80%

Java

60%

Python

90%

c++-logo
C++

70%

pytorch-logo
Pytorch

70%

tensorflow-logo
Tensorflow

60%

keras-logo
Keras

60%

SQL-server-logo
SQL Server

70%

looker-logo
Looker

50%

grafana-logo
Grafana

70%

elastic-logo
Elastic Search

70%

postgreSQL-logo
PostgreSQL

60%

javascript-logo
Javascript

50%

react-logo
React

60%

Experience

 
 
 
 
 
1oggi
Software Engineering
Aug 2021 – Present São Paulo

Responsibilities include:

  • Developed and implemented solutions to track damaged packages in Python, managing to increase the coverage of the number of detected damaged packages
  • Developed and implemented the new damage package declaration service using Javascript and React to reduce the number of damaged packages wrongly.
  • Developed a new functionality using Javascript and React, which adds the creation date of the resolution of the undelivered package, helping to avoid losing the undelivered package within the range of the final delivery date.
  • Development of a new application that is responsible for managing the life cycle of a package from the barcode, developed in NodeJs, Typescript and React.
 
 
 
 
 
Unicamp
Researcher
Unicamp
Mar 2020 – Apr 2024 Campinas

Responsibilities included:

  • Proposed of an image inpainting model leveraging both CNNs and transformers to address challenges posed by large missing regions.
  • Developed of a novel variable hyperparameter strategy applied to the transformer to reduce the computational complexity.
  • Proposed of a image inpainting model by incorporating auxiliary information from the pencil sketch domain, aiming to deal with structural and textural inconsistencies.
 
 
 
 
 
Unicamp and Samsung Electronics America
Researcher
Unicamp and Samsung Electronics America
Apr 2020 – Jun 2021 Campinas

Responsibilities include:

  • Developed new algorithms based on scene representation (LDI and MPI) for the generation of parallax effect motion using a single image, achieving to propose a new light scene representation for restricted scenarios (e.g. cell phones).
  • Implemented and evaluated new algorithms based on GANs and visual transformers for image inpainting, achieving competitive results in comparison with state-of-the-art methods.
 
 
 
 
 
Unicamp and Samsung Electronics America
Researcher
Unicamp and Samsung Electronics America
Aug 2018 – Mar 2020 Campinas

Responsibilities include:

  • Implemented post-processing algorithms to solve problems related to text localization methods using Tesseract OCR, improving the accuracy by 4%.
  • Developed and evaluated new algorithms for multilingual text localization and recognition in images on devices with low computational cost.
 
 
 
 
 
Unicamp
Researcher
Unicamp
Jun 2018 – Jan 2020 Campinas

Responsibilities included:

  • Implemented and developed solutions to fuse text localization results using genetic algorithms (GP), achieving 5% when compared to several baselines.
  • Comaparative study of text localization and recognition approaches for restricted computing scenarios (e.g. mobile devices).
 
 
 
 
 
UNSAAC
Researcher
UNSAAC
Jun 2016 – Jun 2017 Cusco

Responsibilities include:

  • Implemented and evaluated handcrafted and Deep Learning methods to detect and classify hand gesture images under different environments, achieving an accuracy of 96%.
  • Created a new hand gesture dataset based on sign language to train and test our hand gesture detection and classification method.
 
 
 
 
 
Brain Systems
Software Engineering
Jan 2016 – Jun 2017 Cusco

Responsibilities include:

  • Implemented and developed software components to generate XML files and PDF reports of purchases and sales for the electronic invoicing project (BS EFACT), managing to be one of the first companies in the city (Cusco) on generating electronic invoicing in the market.
  • Implemented efficient SQL procedures and views for database queries.

Recent Publications

Quickly discover relevant content by filtering publications.
(2024). Image Inpainting on the Sketch-Pencil Domain with Vision Transformers. 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications.

Cite DOI Link

(2023). Variable-hyperparameter visual transformer for efficient image inpainting. Computers & Graphics.

Cite DOI Link

(2022). Multi-Scale Patch Partitioning for Image Inpainting Based on Visual Transformers. 35th Conference on Graphics, Patterns and Images.

Cite DOI Link

(2021). Adaptive Multiplane Image Generation From a Single Internet Picture. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV).

PDF Cite Link

(2021). Pyramidal Layered Scene Inference with Image Outpainting for Monocular View Synthesis. Computer Analysis of Images and Patterns.

PDF Cite

(2020). Parallax Motion Effect Generation Through Instance Segmentation And Depth Estimation. 2020 IEEE International Conference on Image Processing (ICIP).

PDF Cite Link

(2020). On the Fusion of Text Detection Results: A Genetic Programming Approach. IEEE Access.

PDF Cite DOI Link

(2020). MobText: A Compact Method for Scene Text Localization. Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 5: VISAPP,.

PDF Cite DOI Link

(2019). Pelee-Text: A Tiny Convolutional Neural Network for Multi-oriented Scene Text Detection. 2019 18th IEEE International Conference On Machine Learning And Applications ( ICMLA).

PDF Cite DOI Link

(2019). Multi-Lingual Text Localization via Language-Specific Convolutional Neural Networks. Anais Estendidos da XXXII Conference on Graphics, Patterns and Images.

PDF Cite DOI Link

(2017). Application of convolutional neural networks for static hand gestures recognition under different invariant features. 2017 IEEE XXIV International Conference on Electronics, Electrical Engineering and Computing (INTERCON).

PDF Cite DOI Link

Contact