Behavior Recognition of Moving Objects Using Deep Neural Networks
Jiasong Zhu, Weidong Lin, Ke Sun, Xianxu Hou, Bozhi Liu, Guoping Qiu · 2018
With the rapid development of modern road traffic network, the demand of automatic traffic understanding has become a vital issue for building the intelligent traffic monitoring system and self-driving techniques. In this paper, we focus on behavior recognition of moving objects at busy road intersections in a modern city. To achieve this, we first capture a 4K (3840×2160) traffic video at a busy road intersection of a modern megacity by flying an UAV during the rush hours, and then manually annotate locations and types of road vehicles to form a dataset for this research. Next we propose an innovative behavior recognition framework consists of advanced deep neural network based vehicle detection and localization, type (car, bus and truck) recognition, tracking and behavior recognition over time. We will present experimental results to demonstrate the effectiveness of our solution. This paper not only demonstrates the advantages of using the latest technological advancements (4K video and UAV) but also provides an advanced deep neural network based solution for exploiting these technological advancements for urban traffic analysis.