Simple real-time human detection using a single correlation filter

This paper presents an extremely simple human detection algorithm based on correlating edge magnitude images with a filter. The key is the technology used to train the filter: average of synthetic exact filters (ASEF). The ASEF based detector can process images at over 25 frames per second and achieves a 94.5% detection rate with less than one false detection per frame for sparse crowds. Filter training is also fast, taking only 12 seconds to train the detector on 32 manually annotated images. Evaluation is performed on the PETS 2009 dataset and results are compared to the OpenCV cascade classifier and a state-of-the-art deformable parts based person detector.