Snoop

Olivier Buisson (1) , Jean-Christophe Lombardo (2) , Alexis Joly (3)

1 INA-SUP
2 SED [Sophia] - Service Expérimentation et Développement
3 ZENITH - Scientific Data Management

Snoop is a C++ framework dedicated to large-scale content-based image retrieval. Its main features are (i) the extraction and efficient indexing of visual features (hand-crafted or learned through deep learning), (ii) the search of similar images through approximate k-nearest neighbors and (iii), the supervised recognition of trained visual concepts. The framework can be used either as a set of C++ libraries or as a set of web services through a RESTFUL API. Snoop is the visual search engine used by the Pl@ntNet applications (very large audience).

Domaines: Intelligence artificielle [cs.AI]; Vision par ordinateur et reconnaissance de formes [cs.CV]; Multimédia [cs.MM]; Recherche d'information [cs.IR]; Traitement des images [eess.IV]; Algorithme et structure de données [cs.DS]

Olivier Buisson, Jean-Christophe Lombardo, Alexis Joly. Snoop. 2019. ⟨hal-02096036⟩