.png){kind=icon}
.png){kind=icon}
Return
A multiobject tracking method based on dynamic adaptation and collaborative enhancement
.png){kind=icon}
PRE
AI
DOI：10.1007/s10489-026-07438-0.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Multiobject tracking (MOT) is crucial in areas like intelligent transportation, security, and augmented reality. However, current tracking methods generally fail with different object poses, similar appearances, or complicated motion patterns. A more serious case is in a crowded scene with frequent occlusions that lead to track loss and less accurate tracking. To address these issues, we propose a new MOT approach focused on dynamic adaptation and collaborative enhancement. It consists of a nonlinear adaptive kalman filter that makes adjustments in object states for detecting anomalous motions. We also introduce a nonlinear motion adjustment mechanism that optimizes the prediction step that captures object dynamics. Furthermore, we develop a multi-dimensional feature enhancement network that elaborates on appearance-enhancing exploration accumulations across scales to reduce the input image’s quality effect. To tackle identity switches caused by occlusions, we create a trajectory stitching network that compares the spatiotemporal similarity of track fragments and merges them to maintain track continuity. Our method outperforms other advanced tracking approaches on the MOT17 and MOT20 benchmarks.
Keywords:
Multiobject tracking
Dynamic adaptation
Collaborative enhancement
Journal
IF:
3.5
Papers:
7.5K
Citations:
1.7W