Object detection has a key role in machine learning. Several methods have been proposed for human detection. However, there has not been much effort on the detection of military people. In this paper, at first, we aim to extract the properties of a military person using a rubust HOG detector. Thereafter, we applied an adaboost cascade detector to detect the military person and used some criteria to evaluate the proformance of this method. The results confirm that our porposed method has good capability on the differentiation of military and non-militry people, with little degree of error. Of course, using more training data can increase the proformance of the method. In terms of detection speed, since the method can be applied on autonomous martial robots, the cerelirty in real world applications is acceptable.
