.png){kind=icon}
.png){kind=icon}
Return
Motion estimation for multi-object tracking using KalmanNet with semantic-independent encoding
.png){kind=icon}
PRE
AI
DOI：10.1016/j.inffus.2026.104513.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Propose a novel Semantic-Independent KalmanNet to enhance the accuracy and robustness of motion estimation for existing learning-aided Kalman filters. • Develop a semi-synthetic dataset generation method for training and evaluating learning-aided Kalman filters, enabling their application to multi-object tracking. • Achieve the plug-and-play integration of learning-aided Kalman filters into existing multi-object trackers to enhance their performance. • Release two open-source frameworks, FilterNet and TBDTracker, for benchmarking learning-aided filters and multi-object trackers.
Keywords:
State estimation
Kalman filter
Motion estimation
Multiple-object tracking
Pattern recognition,
Journal
IF:
15.5
Papers:
4.1K
Citations:
2.7W