An Improved Vehicle Tracking Method Based on MDNet

EasyChair Preprint 2360

Abstract

In the field of intelligent transportation, many factors can affect the tracking results of moving vehicles in the video, such as background complexity, illumination change, occlusion, scale transformation and so on. In order to solve the drift problem and improve the tracking accuracy in the vehicle target tracking process, this paper proposed an improved vehicle target tracking algorithm based on MDNet. By combining the instance segmentation method with the MDNet algorithm, the background and vehicle targets can be distinguished remarkably, which enhance tracking performance greatly. The proposed tracking algorithm is evaluated on the OTB dataset. We compared the tracking result of our method with eight mainstream target tracking algorithms. The experimental results illustrate outstanding performance. The target tracking accuracy and tracking success rate of our algorithm achieve good performance in many cases.

Keyphrases: MDNet, instance segmentation, intelligent transportation, target tracking

@booklet{EasyChair:2360,
  author    = {Wang Jianwen and Li Aimin and Pang Yewen},
  title     = {An Improved Vehicle Tracking Method Based on MDNet},
  howpublished = {EasyChair Preprint 2360},
  year      = {EasyChair, 2020}}