Unsupervised Similarity Learning through Cartesian Product of Ranking References for Image Retrieval Tasks

Autor(es)

Valem, Lucas Pascotti ; Guimaraes Pedronette, Daniel Carlos ; IEEE

• Tipo de documento: Artigo Científico
• Fonte: 2016 29TH SIBGRAPI CONFERENCE ON GRAPHICS, PATTERNS AND IMAGES (SIBGRAPI); v. N/A, p. 8-pg., 2016-01-01.

Resumo

Despite the consistent advances in visual features and other Content-Based Image Retrieval techniques, measuring the similarity among images is still a challenging task for effective image retrieval. In this scenario, similarity learning approaches capable of improving the effectiveness of retrieval in an unsupervised way are indispensable. A novel method, called Cartesian Product of Ranking References (CPRR), is proposed with this objective in this paper. The proposed method uses Cartesian product operations based on rank information for exploiting the underlying structure of datasets. Only subsets of ranked lists are required, demanding low computational efforts. An extensive experimental evaluation was conducted considering various aspects, four public datasets and several image features. Besides effectiveness, experiments were also conducted to assess the efficiency of the method, considering parallel and heterogeneous computing on CPU and GPU devices. The proposed method achieved significant effectiveness gains, including competitive state-of-the-art results on popular benchmarks. (AU)

• Processo FAPESP: 14/04220-8 - Execução eficiente de métodos de reclassificação e agregação de listas
• Beneficiário: Lucas Pascotti Valem
• Modalidade de apoio: Bolsas no Brasil - Iniciação Científica
• Processo FAPESP: 13/08645-0 - Reclassificação e agregação de listas para tarefas de recuperação de imagens
• Beneficiário: Daniel Carlos Guimarães Pedronette
• Modalidade de apoio: Auxílio à Pesquisa - Jovens Pesquisadores