Abstract
It focuses on the development of methods that allow image analysis by using
deep neural networks including convolutional neural networks
(CNN), recurrent neural networks (RNN) and visual transforms (ViT).
Among its applications is object recognition, semantic segmentation,
split an image into semantically significant regions, image classification, detection
anomalies and imaging (GAN). These neural networks, though very powerful,
facing the problem of interpretability, so we are also working on
methods that allow these «black boxes» to explain their behavior.
In this context, we have developed works related to the classification of vehicles,
detection of objects, visual detection of graphic anomalies, segmentation of images of
satellite, video-guided autonomous robot programming and image segmentation
medical, among others. We also offer courses focused on Deep Learning and its applications
practices.