Object recognition using geometric properties and a variant of boosting
This paper describes an approach for learning object descriptions as combinations of simple features using labeled still images. The contribution of this paper is a new method for constructing geometric relations of simple features with the LPBoost algorithm. A full search for relevant geometric relations between simple features is rather impossible because of the computation time required. We therefore use a greedy search strategy. In comparison, our recognition results are better than results using related approaches.