Abstract:The evolutionary methods for detection aims to improve the accuracy
and effective of detection task. In this paper, an evolutionary
object detection approach is proposed. The proposed algorithm
employs the 5-fold validation as fitness criterion to direct the
selection of templates, which is vital for creating spatial
histogram features. In evolutionary process, for overcome the
limitation of manually choosing trail vector generation strategies
and its associate control parameters, the self-adaptive evolutionary
method is employed to evolve templates more effectively. We evaluate
the proposed approach compared with two different kinds of methods.
The experimental results prove that by using of self-adapt
evolutionary search, we are able to find object templates with a
higher detection rate.
Lei Cai;Shiru Qu. Evolutionary Spatial Histogram Features for Vehicle Detection[J]. , 2014, 11(11): 3975-3982.
Lei Cai;Shiru Qu. Evolutionary Spatial Histogram Features for Vehicle Detection. , 2014, 11(11): 3975-3982.