Preindication Mining for Predicting Pedestrian Action Change
The action prediction of pedestrians significantly contributes to an intelligent braking system in cars; knowing that the pedestrians will run in several seconds such as for crossing streets, the cars can start braking in advance, to effectively reduce th
- PDF / 3,424,441 Bytes
- 12 Pages / 439.37 x 666.142 pts Page_size
- 80 Downloads / 192 Views
National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1 Umezono, Tsukuba, Ibaraki 305-8568, Japan {kenji.nishida,takumi.kobayashi}@aist.go.jp 2 Mazda Motor Co., 2-5 Moriya-cho Kanagawa-ku, Yokohama, Kanagawa 221-0022, Japan {iwamoto.tar,yamasaki.s}@mazda.co.jp
Abstract. The action prediction of pedestrians significantly contributes to an intelligent braking system in cars; knowing that the pedestrians will run in several seconds such as for crossing streets, the cars can start braking in advance, to effectively reduce the risk for crash accidents. In this paper, we propose a method to predict how the pedestrian act (run or walk) in the future based on preindication in video frames detected by only appearance-based image features. We empirically mine the distinctive frames that precede the target action, ‘running’ in this case, and are effective for predicting it in the framework of feature selection. By using the most effective frames, we can build the action prediction method by exploiting the image features extracted at those frames. As to the image feature extraction methods, we evaluate two types of features in our method, one is GLAC (Gradient Local AutoCorreration) and the other is HOG (Histogram of Oriented Gradient). In the experiments, the effective frames are successfully found around 0.37 s before running action by using GLAC feature; this is not the case of HOG. We also show that the results are closely related to human motion phases from walking to running via biomechanical analysis. Keywords: Action prediction · Feature selection · Intelligent transport system · Image feature extraction
1
Introduction
According to Japanese traffic accident statistics [1], the number of pedestrian accidents are not decreasing while total number of accidents are decreasing. Moreover, the fatality rate in the pedestrian accidents are five times higher than the other accidents. Therefore, prevention of the pedestrian accidents is one of the most urgent issue in our society. The statistics [1] also reports that 70 % of the fatal pedestrian accidents occurred during crossing streets, and thus it is particularly important to safely detect/recognize those crossing pedestrians. c Springer International Publishing AG 2017 J.J. Merelo et al. (eds.), Computational Intelligence, Studies in Computational Intelligence 669, DOI 10.1007/978-3-319-48506-5 18
360
K. Nishida et al.
The fatality risk of pedestrian accidents is actually affected by the impact speed [2]: it is about 4 % at the impact speed of 40 km/h while it increases to about 10 % at 50 km/h and 20 % at 60 km/h. Thus, roughly speaking, the fatality risk decreases by 10 % as the impact speed decreases by 10 km/h. In the situation that automatic emergency braking (AEB) system works on 6 m/s2 as defined by Euro-NCAP [3], it also means that if a car brakes 0.5 s earlier, the fatality risk in pedestrian accidents would be decreased by 10 %. For realizing early braking, it is not sufficient only to detect pedestrians, but it is highly required to recognize
Data Loading...
