Copyright (C) Zenovia Toloudi 2008-2011. This is an experimental project. _ December 15, 2017 18:43:26


PICANICO predicts your architectural taste by associating images of architecture projects to each other according to your preferences.

PICANICO has grown out of the dissertation research of Zenovia Toloudi on architectural identity and taste culture at Harvard Graduate School of Design. PICANICO project started with a machine learning experiment built for Linux environment at 2008 until its recent web-based game version at that was launched at October 2010. PICANICO algorithm and website has been developed by Zenovia Toloudi and George Toloudis [gtroza].

PICANICO gradually "learns" user preferences by classifying their choices in a database of images of buildings. Each image/building is described as a vector of attributes. The vector representation is essentially the brand DNA of the building, since these attributes express the characteristics that users prefer most. PICANICO guesses the architectural taste of the user by proposing similar images based on positive samples ranked by the user. Through this process, PICANICO can offer statistical data for architecture forms and styles as these are linked to personal tastes of the users. Results of PICANICO can be seen in PICANICO blog.

PICANICO etymology: pick + an + icon. [at] gmail [dot] com

back to top