authors picture

Luis Zeni

PhD student at the UFRGS Computer Vision Lab being advised by Professor Claudio Jung. I'm interested in weakly supervised problems in machine learning and deep learning, especially applied to computer vision and object detection applications.

luisfelipezeni ▽ gmail.com - @luis_zeni - /github/ - /blog/

Research

Paper teaser picture
Distilling Knowledge from Refinement in Multiple Instance Detection Networks

Luis Zeni, Claudio Jung

CVPR Deep Vision 2020

In this work, we claim that carefully selecting the aggregation criteria can considerably improve the accuracy of the learned detector. We start by proposing an additional refinement step to an existing approach (OICR), which we call refinement knowledge distillation. Then, we present an adaptive supervision aggregation function that dynamically changes the aggregation criteria for selecting boxes related to one of the ground-truth classes, background, or even ignored during the generation of each refinement module supervision. We call these improvements "Boosted-OICR".

Project page

Paper teaser picture
Real-Time Gender Detection in the Wild Using Deep Neural Networks

Luis Zeni, Claudio Jung

SIBGRAPI 2018

We trained a real-time gender detector in the wild using DNN and visually analyzed which parts leads our model to make mistakes and the bias introduced by the training data.

Project page

Books

Learning to program in python (Aprendendo a programar em python)

EN: This "book" is a series of chapters to help students to learn to program in python (this content is in Portuguese language)

PT: Este livro contempla uma série de capítulos para ajudar estudantes a aprenderem a programar em python

[Ir para o livro]

Teaching

Centro Universitário Ritter dos Reis (2015-2017)

I taught the following courses in the computer science department:

Web Development

Algorithms and Programming

Data Structures

Mobile Development (Android)

Instituto Federal do Rio Grande do Sul (2014)

I taught the following courses in the technical informatics course:

Algorithms and Programming

Mobile Development (Android)

Industry Experience

project picture

*Image Credits: Pixforce

Deep Learning Engineer

Pixforce (Jan 2019 – Mar 2020)

At Pixforce I developed Computer Vision applications using Deep Learning models focused to the industry. I worked in four projects all focused in using computer vision to inspection propouses.

project picture

*Image Credits: Compuletra

Mobile Computer Vision Developer

Compuletra (Jan 2015 – Dec 2015)

I helped in the development of an Android App that uses computer vision to detect car plates and use this information to seek for stolen cars.