ARDOP
This project aims to enhance people observation through robust, discriminative approaches to minimize inaccuracies and address advanced tasks like activity recognition. This enables the development of automatic environment monitoring applications.
DeepEyes
DeepEyes aims to develop algorithmic visual computing and machine intelligence solutions for problems related to forensic computing, digital security and electronic surveillance.
DET
DET proposes the development of methods for pedestrian detection with the aim of reducing the computational cost and maintaining the accuracy obtained by detectors that obtain good results but have a high computational cost.
GigaFrames
GigaFrames main objective is to implement and consolidate an emerging nucleus characterized by the creation of a line of research called Surveillance and Forensic Computing.
HAR-HEALTH
The objective of Har-Health project is the research and development of methods and algorithms capable of automatically recognizing human activities related to chronic diseases (diabetes, hypertension, obesity and aging) based on visual information, signals captured by sensors on personal mobile devices and signals captured by sensors installed in environments.
HITSENSE
Research and development of solutions within the scope of sensor analytics in order to incorporate automatic methods for processing sensory data collected by trackers installed in motor vehicles.
Keymaker
The project addresses the research of new algorithms for recognizing human activities by extracting information such as the importance of objects in carrying out the activity, understanding the interaction of individuals with objects present in the scene.
Literature review on deep learning for the segmentation of seismic images
This review aimed to present a review of Deep Learning approaches for segmenting seismic data through a comprehensive and reproducible review of the literature, identifying and analyzing different approaches, architectures and methodologies used in this field.
Making Sense
One of the main objectives of automatic environmental monitoring is to extract information about activities performed by humans in order to detect interactions between agents and identify patterns of behavior that are suspicious.
SmartView
In order to assist in the monitoring and, consequently, the safety of fans present at sports competitions, this project aims to employ computer vision techniques to automate the resolution of the above problems in order to provide relevant information to those responsible for monitoring fans in sports facilities with the aim of increasing precision and efficiency in decision making.
SMS
This project has two objectives. i) the study, development and evaluation of algorithms to be incorporated into the prototype surveillance system capable of automatically analyzing video and ii) creation, development, implementation and experimental validation of a prototype intelligent surveillance system capable of monitoring workers in regions of the oil exploration platform using visual data.
Subproject 37
Assessment and monitoring of morbidity and mortality in municipalities affected by the rupture of Dam I of the “Córrego do Feijão” Mine
Systematic review of deep learning for structural geological interpretation
This article comprehensively reviews current techniques, models, and practices in DL-based seismic volume interpretation based on three different structural interpretation tasks: Fault interpretation, Horizon estimation, and Relative Geological Time (RGT) estimation, three complementary pillars of the geological framework.
V+
In this project, visual data captured from environmental monitoring cameras will be used as input for analysis modules (solutions based on computer vision and machine learning to be developed). All solutions will be implemented on embedded devices with little computational power and, therefore, requiring research aimed at compressing deep models.
VER+
This project focuses on solving problems related to large-scale visual surveillance, where data is acquired from multiple surveillance cameras.