BOOSTING : FROM DATA TO HARDWARE USING AUTOMATIC IMPLEMENTATION TOOL (ThuAmPO2)

Author(s) : Johel Miteran (le2i, France) Jiri Matas (CMP, Czech Republic) Julien Dubois (Le2i, France) Elbey Bourennane (Le2i, france)

Abstract : We propose a method of automatic hardware imple-mentation of a decision rule based on the Adaboost algo-rithm. We review the principles of the classification method and we evaluate its hardware implementation cost in term of FPGA’s slice, using different weak classi-fiers based on the general concept of hyperrectangle. We show how to combine the weak classifiers in order to find an efficient trade-off between classification per-formances and hardware implementation cost. We pre-sent results obtained using examples coming from UCI databases.

Menu