Publicação

Face-based Photo Indexing in Edge Computing Environments

Ver documento

Detalhes bibliográficos
Resumo:Over recent years, smart mobile devices have grown in popularity. With such popularity growth, data traffic has also increased, which gave rise to new problems such as higher latency in data requests or less data storage capability. A paradigm that promises to nullify many of the issues with this growth is Edge Computing. A computing paradigm composed by the user devices, edge servers and the cloud. Edge Servers, located at the edge of the internet, close to the user’s devices, help processing and disseminating information and data. During this dissertation, we propose to enhance Chives, a machine learning API to identify faces in photos, in this environment, creating a cluster indexing system, in order to enable the development of a photo sharing application with lower latency in picture search through facial recognition. Such index is created using Conflict-free Replicated Data Types. On a mobile phone, the user will be able to search for photos with faces, using a photo of a similar face as an input. Our solution proved to retrieve the correct results, while being almost as fast as the human eye perception’s capability, being a good addition to the EdgeGarden environ- ment.
Autores principais:Pregal, Filipe Lourenço Lopes
Assunto:Conflict-Free Replicated Data Type Edge Computing Indexing at the Edge Machine Learning Publish/Subscribe
Ano:2024
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:Over recent years, smart mobile devices have grown in popularity. With such popularity growth, data traffic has also increased, which gave rise to new problems such as higher latency in data requests or less data storage capability. A paradigm that promises to nullify many of the issues with this growth is Edge Computing. A computing paradigm composed by the user devices, edge servers and the cloud. Edge Servers, located at the edge of the internet, close to the user’s devices, help processing and disseminating information and data. During this dissertation, we propose to enhance Chives, a machine learning API to identify faces in photos, in this environment, creating a cluster indexing system, in order to enable the development of a photo sharing application with lower latency in picture search through facial recognition. Such index is created using Conflict-free Replicated Data Types. On a mobile phone, the user will be able to search for photos with faces, using a photo of a similar face as an input. Our solution proved to retrieve the correct results, while being almost as fast as the human eye perception’s capability, being a good addition to the EdgeGarden environ- ment.