Multi-view Tracking Using Weakly Supervised Human Motion Prediction
It addresses occlusion handling in crowded scenes for surveillance or robotics applications, presenting an incremental improvement over existing tracking-by-detection methods.
The paper tackles multi-view people tracking by predicting human motion over time and inferring presence from these predictions, outperforming state-of-the-art methods on PETS2009 and WILDTRACK datasets.
Multi-view approaches to people-tracking have the potential to better handle occlusions than single-view ones in crowded scenes. They often rely on the tracking-by-detection paradigm, which involves detecting people first and then connecting the detections. In this paper, we argue that an even more effective approach is to predict people motion over time and infer people's presence in individual frames from these. This enables to enforce consistency both over time and across views of a single temporal frame. We validate our approach on the PETS2009 and WILDTRACK datasets and demonstrate that it outperforms state-of-the-art methods.