References & Citations
Computer Science > Computer Vision and Pattern Recognition
Title: Who Make Drivers Stop? Towards Driver-centric Risk Assessment: Risk Object Identification via Causal Inference
(Submitted on 5 Mar 2020 (v1), last revised 3 Aug 2020 (this version, v2))
Abstract: A significant amount of people die in road accidents due to driver errors. To reduce fatalities, developing intelligent driving systems assisting drivers to identify potential risks is in an urgent need. Risky situations are generally defined based on collision prediction in the existing works. However, collision is only a source of potential risks, and a more generic definition is required. In this work, we propose a novel driver-centric definition of risk, i.e., objects influencing drivers' behavior are risky. A new task called risk object identification is introduced. We formulate the task as the cause-effect problem and present a novel two-stage risk object identification framework based on causal inference with the proposed object-level manipulable driving model. We demonstrate favorable performance on risk object identification compared with strong baselines on the Honda Research Institute Driving Dataset (HDD). Our framework achieves a substantial average performance boost over a strong baseline by 7.5%.
Submission history
From: Chengxi Li [view email][v1] Thu, 5 Mar 2020 04:14:35 GMT (4882kb,D)
[v2] Mon, 3 Aug 2020 04:28:21 GMT (9443kb,D)
Link back to: arXiv, form interface, contact.