A multi-object tracker using dynamic Bayesian networks and a residual neural network based similarity estimator

作者：

Highlights：

• A new multi-object tracker using a dynamic Bayesian network is introduced

• State of the art residual neural network for extracting feature descriptors

• DBN trained on MOTChallenge data and Feature extractor on the MARS pedestrian dataset

• IoU and similarity distances along with Hungarian algorithm for max cost assignment

• A custom trained object detector for analysing effects on tracker performance

摘要

•A new multi-object tracker using a dynamic Bayesian network is introduced•State of the art residual neural network for extracting feature descriptors•DBN trained on MOTChallenge data and Feature extractor on the MARS pedestrian dataset•IoU and similarity distances along with Hungarian algorithm for max cost assignment•A custom trained object detector for analysing effects on tracker performance

论文关键词：Multi-object tracking,Dynamic Bayesian networks,Residual neural networks,YOLO V5,MOTChallenge

论文评审过程：Received 27 July 2021, Revised 11 August 2022, Accepted 26 September 2022, Available online 30 September 2022, Version of Record 21 October 2022.

论文官网地址：https://doi.org/10.1016/j.cviu.2022.103569