Mono3D++: Monocular 3D Vehicle Detection with Two-Scale 3D Hypotheses and Task Priors

Abstract

We present a method to infer 3D pose and shape of vehicles from a single image. To tackle this ill-posed problem, we ensure two-scale projection consistency between the generated 3D hypotheses and their 2D pseudo-measurements. Specifically, we use a morphable wireframe model to generate a finescaled and accurate representation of vehicle shape and pose. To reduce its sensitivity to 2D landmarks, we jointly model the 3D bounding box as a coarse representation and improve the robustness. We also integrate three task priors, including unsupervised monocular depth, a ground plane constraint as well as vehicle shape priors, with forward projection errors into an overall energy function. Our system is benchmarked on the KITTI dataset and outperforms state-of-the-art methods for monocular 3D vehicle detection.

References

  • Mono3D++: Monocular 3D Vehicle Detection with Two-Scale 3D Hypotheses and Task Priors.
    Tong He, Stefano Soatto. AAAI Conference on Artificial Intelligence (AAAI), 2019. [PDF]