Framework improves key object detection from driver perspective
A Two-Stage Framework for Ego-Centric Key Object Identification via Object State Prediction
Computer Vision and Pattern Recognition
Summary
In self-driving cars, it's important to know which objects around the car really matter. The existing ways mostly look at objects on their own or how objects relate visually, but they don't focus on what matters most from the car's own viewpoint. The authors designed a two-step method that first guesses how objects are likely to behave relative to the car, then uses that information along with where the objects are and how they move to decide which objects need attention. Their tests on real driving data show this approach helps in spotting the most important objects in tricky traffic situations.
autonomous drivingego-vehicle perspectiveobject detectionobject state predictionspatial-temporal reasoningkey object identificationcomputer visiontraffic environmentbehavior estimation
Authors
Shihong Ling, Yue Wan, Xiaowei Jia, Na Du
Abstract
This paper presents a novel framework designed to enhance key object identification in autonomous driving. Existing methods primarily focus on either detecting objects independently or leveraging visual relationships, but they do not explicitly consider the ego vehicle's perspective in determining object importance. To address this gap, we propose a structured approach that integrates a virtual ego-vehicle representation and a modular object state predictor, enabling a more accurate estimation of object behaviors relative to the ego-vehicle. Subsequently, our framework employs spatial-temporal reasoning to refine key object identification, prioritizing objects based on their states and relative spatial information rather than relying solely on visual relationships. Experimental results on real-world driving datasets demonstrate the effectiveness of our approach in accurately detecting critical objects in complex traffic environments.