Storing Feature Vectors in Relational Image Data Warehouses

Autor(es):Mostrar menos -

Rocha, Guilherme Muzzi ; Capelo, Piero Lima ; Carniel, Anderson Chaves ; Aguiar, Cristina Dutra ; Chiusano, S ; Cerquitelli, T ; Wrembel, R ; Norvag, K ; Catania, B ; Vargas-Solar, G ; Zumpano, E

• Tipo de documento: Artigo Científico
• Fonte: NEW TRENDS IN DATABASE AND INFORMATION SYSTEMS, ADBIS 2022; v. 1652, p. 9-pg., 2022-01-01.

Resumo

An image data warehousing also manipulates image data represented by feature vectors and attributes for similarity search. In this paper, we study the impact of storing feature vectors on the performance of analytical queries extended with a similarity search predicate over images. We consider the management of huge data volumes. Thus, we use Spark as support. Experimental results showed that depending on the query characteristics and the data warehouse design, the difference in performance was up to 86.23%. Based on these results, we propose guidelines for storing feature vectors in relational image data warehouses. (AU)

• Processo FAPESP: 18/10607-3 - Processamento de consultas analíticas em ambientes computacionais paralelos e distribuídos
• Beneficiário: Guilherme Muzzi da Rocha
• Modalidade de apoio: Bolsas no Brasil - Mestrado
• Processo FAPESP: 18/22277-8 - Processamento de Consultas OLAP e SOLAP em Ambientes Computacionais Paralelos e Distribuídos
• Beneficiário: Cristina Dutra de Aguiar
• Modalidade de apoio: Auxílio à Pesquisa - Regular