Offline Tracking with Object Permanence

arXiv:2310.01288v4 Announce Type: replace
Abstract: To reduce the expensive labor cost for manual labeling autonomous driving datasets, an alternative is to automatically label the datasets using an offline perception system. However, objects might be temporally occluded. Such occlusion scenarios in the datasets are common yet underexplored in offline auto labeling. In this work, we propose an offline tracking model that focuses on occluded object tracks. It leverages the concept of object permanence which means objects continue to exist even if …

abstract alternative arxiv auto autonomous autonomous driving cost cs.cv datasets driving however labeling labor object objects offline perception reduce tracking type work