Just Go with the Flow: Self-Supervised Scene Flow Estimation
This addresses the need for accurate motion tracking of multiple objects in autonomous driving without requiring costly annotated data, representing an incremental improvement over existing methods.
The paper tackles the problem of scene flow estimation for autonomous driving by proposing a self-supervised training method using nearest neighbors and cycle consistency losses, which matches state-of-the-art supervised performance without real-world annotations and exceeds it when combined with a smaller labeled dataset.
When interacting with highly dynamic environments, scene flow allows autonomous systems to reason about the non-rigid motion of multiple independent objects. This is of particular interest in the field of autonomous driving, in which many cars, people, bicycles, and other objects need to be accurately tracked. Current state-of-the-art methods require annotated scene flow data from autonomous driving scenes to train scene flow networks with supervised learning. As an alternative, we present a method of training scene flow that uses two self-supervised losses, based on nearest neighbors and cycle consistency. These self-supervised losses allow us to train our method on large unlabeled autonomous driving datasets; the resulting method matches current state-of-the-art supervised performance using no real world annotations and exceeds state-of-the-art performance when combining our self-supervised approach with supervised learning on a smaller labeled dataset.