Gabriel Bertocco

Gabriel Bertocco

Ph.D. candidate in Computer Science and member of the Recod.ai lab

University of Campinas

Recod.ai

Biography

I am currently a Ph.D. student at University of Campinas (UNICAMP), Brazil. I have expertise in Deep Learning, Self-Supervised Learning and Computer Vision. In my current Ph.D. work I study Self-Supervised Learning and Clustering techniques to handle large unlabeled datasets tackling biases in different image semantics (People, Objects and Places) in the context of Re-Identification, and modalities (images and texts). Besides I am Teacher Assistant on Complex Data Mining promoted by Escola de Extensão da UNICAMP. During my academic life I have published in top-tier venues, being the my last work entitled “Unsupervised and Self-Adaptative Techniques for Cross-Domain Person Re-Identification” the most important one thus far of my Ph.D, published in the prestigious IEEE Transactions on Information Forensics and Security (T-IFS) with impact factor of 7.178.

Interests
  • Computer Vision
  • Deep Learning
  • Self-Supervised Learning
  • Artificial Intelligence
Education
  • Ph.D. in Computer Science focused in Computer Vision and Deep Learning, Ongoing

    University of Campinas - Institute of Computing

  • BSc in Compiter Engineering, 2019

    University of Campinas

Publications

Quickly discover relevant content by filtering publications.
(2022). Reasoning for Complex Data through Ensemble-based Self-Supervised Learning. ArXiv.

DOI

(2021). Cross-dataset emotion recognition from facial expressions through convolutional neural networks. Journal of Visual Communication and Image Representation.

DOI

(2021). Unsupervised and self-adaptative techniques for cross-domain person re-identification. IEEE Transactions on Information Forensics and Security.

DOI

(2021). The Artificial Intelligence and the challenges on the Digital Forensics Science in the XXI century.

DOI

(2020). Forensic event analysis: From seemingly unrelated data to understanding. IEEE Security and Privacy.

DOI

(2020). Two-tiered face verification with low-memory footprint for mobile devices. IET Biometrics.

DOI

(2020). Detecting face presentation attacks in mobile devices with a patch-based CNN and a sensor-aware loss function. PloS One.

DOI

(2019). Deep face verification for spherical images. 2019 IEEE International Conference on Image Processing (ICIP).

DOI

(2017). Automatic age range estimation on mobile devices. XV Congress of Scientific Initiation at UNICAMP.

DOI

Experience

 
 
 
 
 
Ph.D. student
Institute of Computing/UNICAMP
Aug 2019 – Present Campinas-SP
This is my current project that I have focused on. The ultimate goal is to design large-scale self-supervised learning solutions to filter and group persons, objects and places in fully-unsupervised manner avoiding biases. We envision to design solutions that can also be extended for applications in further Artificial Intelligence research fields.
 
 
 
 
 
Undergraduate Research
Scipet Solutions
Jan 2019 – Jul 2019 Campinas-SP, Brazil
I worked in deep learning-based solutions to identify stray dogs and cats and filter out unknown classes leveraging and studying Open-Set solutions and Explainable Artificial Intelligence (XAI) techniques. Research outcomes are now available, in the form of mobile applications, to some prefectures in Brazil
 
 
 
 
 
Undergraduate Research
Motorola Mobility
Sep 2015 – Dec 2018 Campinas-SP, Brazil
In this projected I explored Deep Learning-based solutions to estimate the age range of a given a photo of a person taken by a mobile device. This work has been appreciated as the best undergraduate project on University of Campinas in 2017.
 
 
 
 
 
Undergraduate Student
University of Campinas (UNICAMP)
Mar 2014 – Jun 2019 Campinas-SP, Brazil

Patents

(2018). Multiple-tiered facial recognition. United States.

Projects

*

Contact