We find shadows in many images and videos. Traditionally, shadows are considered as noises because they make\nhurdles for visual tasks such as detection and tracking. In this work, we show that shadows are helpful in pedestrian\ndetection instead. Occlusions make pedestrian detection difficult. Existing shape-based detection methods can have\nfalse-positives on shadows since they have similar shapes with foreground objects. Appearance-based detection\nmethods cannot detect heavily occluded pedestrians. To deal with these problems, we use appearance, shadow, and\nmotion information simultaneously in our method. We detect pedestrians using appearance information of\npedestrians and shape information of shadow regions. Then, we filter the detection results based on motion\ninformation if available. The proposed method gives low false-positives due to the integration of different features.\nMoreover, it alleviates the problem brought by occlusions since shadows can still be observable when foreground\nobjects are occluded. Our experimental results show that the proposed algorithm provides good performance in\nmany difficult scenarios.
Loading....