Abstract
We propose an end-to-end network that takes a single perspective RGB image of a complex road scene as input, to produce occlusion-reasoned layouts in perspective space as well as a top-view parametric space. In contrast to prior works that require dense supervision such as semantic labels in perspective view, the only human annotations required by our method are for parametric attributes that are cheaper and less ambiguous to obtain. To solve this challenging task, our design is comprised of modules that incorporate inductive biases to learn occlusion-reasoning, geometric transformation and semantic abstraction, where each module may be supervised by appropriately transforming the parametric annotations. We demonstrate how our design choices and proposed deep supervision help achieve accurate predictions and meaningful representations. We validate our approach on two public datasets, KITTI and NuScenes, to achieve state-of-the-art results with considerably lower human supervision.
Abstract (translated)
URL
https://arxiv.org/abs/2104.06730
