Abstract: Open-vocabulary multiple object tracking (MOT) aims to track arbitrary objects in the real world. Although significant progress has been achieved in object classification by leveraging the ...
Our proposed HOMATracker comprises the following steps: (A) Object detection: A detector is used to locate the objects in each frame; (B) Instance-level appearance feature extraction: We involve a ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results