Large-Scale Coarse-to-Fine Object Retrieval Ontology and Deep Local Multitask Learning

Figure 5

(a) Extracting regions, categories, and attributes from a query image with trained models of the CFOR system. After that, users can use this semantic information to reduce the searching space. (b) Fashion ontology used to retrieve. (c) Retrieval results.

(a)

(b)

(c)