Afjpda: a multiclass multi-object tracking with appearance feature-aided joint probabilistic data association

HIGHLIGHTS

• What: To alleviate issues with clutters the authors propose the appearance feature-aided joint probabilistic data association filter. The authors propose a tracking filter that is directly applicable for multiclass multi-object tracking and significantly reduces fragmentation and ID switch by using a JPDA filter and appearance feature. The authors propose a new approach, called appearance feature-aided JPDA (AFJPDA), which uses a fusion of Mahalanobis distance and cosine distance between objects in image coordinates to improve the performance of association after occlusion or missed detection. The authors propose an adaptive gating logic that can change the . . .
  

   

  

  If you want to have access to all the content you need to log in!

  Thanks :)

  Username or Email

  Password

  Remember me

  Lost your password?

  If you don't have an account, you can create one here.