Feature and Box Propagation for Video Vehicle Detection
Yanni Yang, Huansheng Song, Zhe Dai, Wentao Zhang, Yan Chen · CICTP 2020 · 2020
Video vehicle detection is more valuable and challenging than image vehicle detection for an intelligent transportation system. Due to the existing situation of vehicle blurring, occlusion, and scale changing in traffic monitoring, using static vehicle detection network often leads to the decrease of detection accuracy. In this paper, based on DFF method, we fuse a tracking algorithm to realize box propagation, and form a new video vehicle detection framework that considers both detection accuracy and speed. We extract feature maps in key frames, and propagate feature maps and boxes in non-key frames. Compared with static detectors, the proposed method greatly improves the consistency of video vehicle detection results. In addition, the detection performance of our method is obviously superior to the basic detector DFF in accuracy.