Publicação

Infrastructure for machine learning and computer vision

Ver documento

Detalhes bibliográficos
Resumo:The infrastructure surrounding machine learning projects is of utmost importance: Machine learning projects require data acquisition mechanisms, software for data processing, as well as a benchmarking platform for evaluating performance of machine learning algorithms over time. In this report we describe our work aimed at developing such infrastructure for a Europe based computer vision startup specializing in human behaviour tracking. We discuss three projects comprising the work. One dedicated to creating a machine learning dataset for human behaviour monitoring, another to developing a screen-camera calibration tool, and third to setting up a benchmarking platform. The projects were integrated with the core technology of the startup, and will continue to be applied in the future.
Autores principais:Chikhladze, Dimitri
Assunto:Data acquisition mechanisms Data processing Benchmarking platform Human behaviour monitoring
Ano:2019
País:Portugal
Tipo de documento:dissertação de mestrado
Tipo de acesso:acesso aberto
Instituição associada:Universidade Nova de Lisboa
Idioma:inglês
Origem:Repositório Institucional da UNL
Descrição
Resumo:The infrastructure surrounding machine learning projects is of utmost importance: Machine learning projects require data acquisition mechanisms, software for data processing, as well as a benchmarking platform for evaluating performance of machine learning algorithms over time. In this report we describe our work aimed at developing such infrastructure for a Europe based computer vision startup specializing in human behaviour tracking. We discuss three projects comprising the work. One dedicated to creating a machine learning dataset for human behaviour monitoring, another to developing a screen-camera calibration tool, and third to setting up a benchmarking platform. The projects were integrated with the core technology of the startup, and will continue to be applied in the future.