Object detection by joint features based on two-stage boosting

In this paper, we propose an object detection method that uses Joint features combined from multiple Histograms of Oriented Gradients (HOG) feature using two-stage boosting. There has been much research in recent years on statistical training methods and object detection methods that combine low-level features obtained from local areas. In our approach, multiple low-level HOG features are combined by using Real AdaBoost to automatically generate Joint features. Joint features represent the co-occurrence of the HOG features of multiple cells combined by the first-stage Real AdaBoost. Next, the generated Joint features pool is input to the second-stage Real AdaBoost, which constructs the final classifier. In this way, it is possible to capture shape symmetry and edge continuity, which single HOG features cannot do, so highly accurate detection is possible. In this paper we report experiments involving the detection of humans and vehicles performed to test the effectiveness of the proposed method. In addition, two-stage boosting classifier is used to represent the co-occurrence of the appearance(HOG) and spatiotemporal(PSA) features for detecting pedestrian. The use of feature co-occurrence, which captures the similarity of appearance, motion, and spatial information within the human class, makes it an effective detector.