Bagging-Adaboost Ensemble with Genetic Algorithm Post Optimization for Object Detection

We propose a novel learning algorithm, called Bagging-Adaboost ensemble algorithm with genetic algorithm post optimization, for object detection that uses local shape-based feature. The feature is motivated by the scheme that use the chamfer distance as a shape comparison measure. It can be calculated very quickly using a look-up table. Random sampling boosting algorithm is used to select a discriminative edge shape features set from a over-complete dictionary of features and form an object detector. Genetic algorithm post optimization procedure is used to remove based classifiers which cause higher error rates. The resulting classifier consists of fewer base classifiers yet achieves better generalization performance. To demonstrate our method we trained a system to detect pedestrians in complex natural scenes. Experimental results show that our system can extremely rapidly detect objects with high detection rate. The result is very competitive with other published object detection schemes. The learning techniques can be extended to detect other objects.