News

We want to congratulate now Doctor Jefferson Fontinele for finishing his Ph.D. Doctor Fontinele researched semantic segmentation deep learning methods on traffic and biomedical images. Two works from our lab (“Faster α-expansion via dynamic programming and image partitioning” and “Attention-based fusion of semantic boundary and non-boundary information to improve semantic segmentation”) are direct outcomes of his thesis entitled “Paying attention to the boundaries in semantic image segmentation,” confirming its success in the field of semantic segmentation. Well done, Doctor Jefferson Fontinele!

We are glad to announce that our colleagues from IvisionLab had two papers accepted in the 17th International Symposium on Medical Information Processing and Analysis (SIPAIM). Laís Pinheiro, Bernardo Silva, Brenda Sobrinho, Fernanda Lima, Patrícia Cury, and Luciano Oliveira from the OdontoAI project submitted an article entitled Numbering permanent and deciduous teeth via deep instance segmentation in panoramic X-rays. In their work, they detail instance segmentation experiments on a soon-to-be public data set. Paulo Chagas, Luiz Souza, Rodrigo Calumby, Izabelle Pontes, Stanley Araújo, Angelo Duarte, Nathanael Pinheiro, Washington Santos, and Luciano Oliveira had their work entitled Toward unbounded open-set recognition to say “I don’t know” for glomerular multi-lesion classification accepted for publication. The former paper is from the PathoSpotter project, in which the IvisionLab is highly active.

We are glad to announce that our colleagues from IvisionLab, Gabriel Lefundes, and Luciano Oliveira, had a paper accepted in the 2021 Conference on Graphics, Patterns and Images (SIBGRAPI). The work is entitled “Gaze estimation via self-attention augmented convolutions” and describes a novel neural network architecture, the ARes-gaze (Attention-augmented ResNet). The work is part of Ivision’s research on biometric systems, specifically on gaze estimation.

We are pleased to announce another IvisionLab publication. Our colleagues Joao Barros and Luciano Oliveira just had a paper accepted in IEEE Intelligent Vehicles Symposium. The work is entitled “Deep speed estimation from synthetic and monocular data” and describes a solution for estimating the speed of vehicles with monocular cameras.

We are glad to announce that our colleagues from IvisionLab had two papers accepted in the 2021 Brazilian Symposium on Computing Applied to Health (SBCAS). Paulo Chagas, Luiz Souza, Rodrigo Calumby, Angelo Duarte, Washington L.C. dos-Santos and Luciano Oliveira published an article entitled Deep-learning-based membranous nephropathy classification and Monte-Carlo dropout uncertainty estimation, while Sarah Cerqueira, Ellen Aguiar, Angelo Duarte, Washington dos Santos, Luciano Oliveira and Michele Angelo published a paper entitle PathoSpotter Classifier: Um Serviço Web para Auxílio à Classificação de Lesões em Glomérulos Renais. Both papers are from the PathoSpotter project, in which the IvisionLab is highly active.