POKER-DVS Database

and

Instituto de Microelectrónica de Sevilla, IMSE-CNM (CSIC & Univ. of Sevilla), Sevilla, SPAIN.

The POKER-DVS database consists of a set of 131 poker pip symbols tracked and extracted from 3 separate DVS recordings, while browsing very quickly poker cards.

A first version with only 40 symbols was used in the ConvNet symbol recognition paper:

J. A. Pérez-Carrasco, et. al, "Mapping from Frame-Driven to Frame-Free Event-Driven Vision Systems by Low-Rate Rate-Coding and Coincidence Processing. Application to Feed-Forward ConvNets," IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 35, No. 11, pp. 2706-2719, Nov. 2013. (video). Doi: http://doi.ieeecomputersociety.org/10.1109/TPAMI.2013.71. See also this nice video provided by T. Delbruck (click).

The DVS used is a high-sensitive DVS described in:

T. Serrano-Gotarredona and B. Linares-Barranco, "A 128x128 1.5% Contrast Sensitivity 0.9% FPN 3us Latency 4mW Asynchronous Frame-Free Dynamic Vision Sensor Using Transimpedance Amplifiers," IEEE J. Solid-State Circuits, vol.48, No. 3, pp. 827-838, March 2013 (available from:

http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6407468).

Original DVS recorded files are 128x128 pixel and can be displayed with jAER (available from: http://sourceforge.net/p/jaer/wiki/Home). Matlab^TM scripts are provided to track and extract individual poker pip symbols as they cross the screen and extract them as 32x32 pixel event streams, each of which can also be visualized with jAER. The extracted symbols are readily available (no need to run the Matlab^TM scripts).

This database can be downloaded free of charge, provided it is used for non-commercial purposes and the original source is credited in publications and reports.

To download it go to http://imse-cnm.csic.es/caviar/POKER_DVS.