Learning to Detect Mobile Objects from LiDAR Scans Without Labels
This enables cheap and scalable object detection for autonomous driving by eliminating the need for costly labeled data, though it is incremental in leveraging existing self-training methods.
The paper tackles the problem of training 3D object detectors for autonomous driving without human-annotated labels by using unlabeled LiDAR data and common-sense heuristics to generate seed labels, achieving surprisingly accurate detection through self-training.
Current 3D object detectors for autonomous driving are almost entirely trained on human-annotated data. Although of high quality, the generation of such data is laborious and costly, restricting them to a few specific locations and object types. This paper proposes an alternative approach entirely based on unlabeled data, which can be collected cheaply and in abundance almost everywhere on earth. Our approach leverages several simple common sense heuristics to create an initial set of approximate seed labels. For example, relevant traffic participants are generally not persistent across multiple traversals of the same route, do not fly, and are never under ground. We demonstrate that these seed labels are highly effective to bootstrap a surprisingly accurate detector through repeated self-training without a single human annotated label.