Heitor Rapela Medeiros

I am a fourth-year PhD Candidate at ÉTS Montreal, where I work in the Distech Controls Industrial Research Chair project with an MITACS scholarship. I am advised by Professors Marco Pedersoli and Eric Granger. My primary research focus is computer vision and machine learning, especially in the areas of object detection and domain adaptation. I am also a member of the International Laboratory on Learning Systems (ILLS).

I did my Masters degree at CIn-UFPE, advised by Professor Hansenclever Bassani . In parallel, I worked as a researcher in LINCS/CETENE with Professor Edna Barros and Antonyus Pyetro in the field of object detection and face recognition. By the end of my Masters, I worked as a junior data engineer at Neurotech for eight months, where I did ETL, data ingestion and implemented deep learning solutions.

Regarding my undergraduate studies, I accomplished my bachelor's in computer engineering at CIn-UFPE, where I had the opportunity to work in robotics with Professors Edna Barros and Hansenclever Bassani and in computer vision with Professor Carlos Alexandre. During my Bachelor's, I did an internship as a software engineer at Motorola/CIn, where I developed a parser solution for 4G wireless communication, and I co-founded RoboCIn, the best robot competition team in Latin America. Over the course of my academic career, I have been awarded several scholarships and have won hackathons and robotics competitions.

I am a volunteer reviewer for WACV, ICLR, and LXAI, where I currently serve as co-chair for the LXAI NeurIPS 2024 workshop. In my free time, I enjoy participating in Kaggle competitions, playing guitar, watching anime, and drinking a good cappuccino.

I am currently seeking internship positions for 2025. Upon completing my PhD, I will explore options for the next stage of my career, whether in a postdoctoral position or in the industry. Please feel free to contact me if you have any opportunities.

CV | Google Scholar | GitHub | Twitter | LinkedIn
heitor.rapela-medeiros.1 [at] ens.etsmtl.ca
hrm [at] cin.ufpe.br

Currently, my main interests include object detection, generative models,
domain adaptation, diffusion models, parameter efficient fine-tuning and visual prompt tuning.

[WACV 2025] MiPa: Mixed Patch Infrared-Visible Modality Agnostic Object Detection
Heitor R. Medeiros*, D Latortue*, E Granger, M Pedersoli
Winter Conference on Applications of Computer Vision, 2025.
Early accepted in the first round.
arxiv | code | * After we will provide more info.

[ECCV 2024] Modality Translation for Object Detection Adaptation Without Forgetting Prior Knowledge
Heitor R. Medeiros, Masih Aminbeidokhti, Fidel Guerrero Pena, David Latortue, Eric Granger, Marco Pedersoli
European Conference on Computer Vision, Milan Italy, 2024.
arXiv | code | tweet

[WACV 2024] HalluciDet: Hallucinating RGB Modality for Person Detection Through Privileged Information
Heitor R. Medeiros, Fidel A. Guerrero Pena, Masih Aminbeidokhti, Thomas Dubail, Eric Granger, Marco Pedersoli
Winter Conference on Applications of Computer Vision, 2024.
website | code | video | LIVIA Presentation | poster | tweet

[WACV 2024] Domain Generalization by Rejecting Extreme Augmentations
Masih Aminbeidokhti, Fidel A. Guerrero Peña, Heitor R. Medeiros, Thomas Dubail, Eric Granger, Marco Pedersoli
Winter Conference on Applications of Computer Vision, 2024.
arXiv | code | LIVIA Presentation

[CVPR 2023] Re-Basin via Implicit Sinkhorn Differentiation
Fidel A. Guerrero Peña, Heitor R. Medeiros, Thomas Dubail, Masih Aminbeidokhti, Eric Granger, Marco Pedersoli
Conference on Computer Vision and Pattern Recognition, 2023
website | paper | code

[ECCV 2022 RWS Workshop] Privacy-Preserving Person Detection Using Low-Resolution Infrared Cameras
Thomas Dubail, Fidel A. G. Peña, Heitor R. Medeiros, Masih Aminbeidokhti, Eric Granger, Marco Pedersoli
European Conference on Computer Vision, 2022 Real-World Surveillance: Applications and Challenges Workshop, 2022
springer | arXiv | code

[IJCNN 2020] Deep Categorization with Semi-Supervised Self-Organizing Maps
Pedro H. M. Braga, Heitor R. Medeiros , Hansenclever F. Bassani International Joint Conference on Neural Networks, 2020
paper

[arXiv 2020] Learning to Play Soccer by Reinforcement and Applying Sim-to-Real to Compete in the Real World
Hansenclever F Bassani, Renie A Delgado, José Nilton de O Lima Junior, Heitor R. Medeiros, Pedro HM Braga, Alain Tapp
arXiv preprint arXiv:2003.11102, 2020
arXiv

[DAES 2020] An embedded automatic license plate recognition system using deep learning
Diogo MF Izidio, Antonyus PA Ferreira, Heitor R Medeiros, Edna N da S Barros
Design Automation for Embedded Systems, 2020
paper

[CVIU 2019] Dynamic topology and relevance learning SOM-based algorithm for image clustering tasks
Heitor R Medeiros, Felipe DB de Oliveira, Hansenclever F Bassani, Aluizio FR Araujo Computer Vision and Image Understanding, 2019
paper

[GECCO 2019] Latin hypercube initialization strategy for design space exploration of deep neural network architectures
Heitor R Medeiros, Diogo MF Izidio, Antonyus P do A Ferreira, Edna N da S. Barros
Genetic and Evolutionary Computation Conference Companion, 2019
paper

[IJCNN 2018] SegNetRes-CRF: A deep convolutional encoder-decoder architecture for semantic image segmentation
Luiz Antônio de Oliveira Junior, Heitor R Medeiros, David Macêdo, Cleber Zanchettin, Adriano LI Oliveira, Teresa Ludermir International joint conference on neural networks, 2018
springer