Marker Based Pedestrian Detection Using Augmented Reality
Chanyong Kim, Hyo-Jin Kim, Huiseung Son, Muhammad Tanseef Shahid, Hyun Chul Song, Francesco Piccialli, Kwang Nam Choi · 2019
Pedestrian detection is a popular research topic from the last decade. Most of the pedestrian identification models are based on face recognition algorithms. It is a difficult task to detect and track individual pedestrians based on these algorithms because there is a compulsion that their faces should be towards the camera. This limitation makes face recognition algorithms inefficient to detect pedestrians from the backsides. In this paper, we proposed a method for pedestrian detection using marker recognition. Multiple pedestrians are detected and then tracked based on markers attached to their backsides. Attaching markers on back-side of pedestrians helps to recognize them even when they are looking the other way. After the marker is recognized, the unique character related to that marker is displayed as a 3D object. This marker-based pedestrian detection is carried out using a mobile phone system and can be applied to embedded systems. The proposed method makes it possible to recognize up to three pedestrians located at different positions from the camera.