Deep Learning for Vanishing Point Detection Using an Inverse Gnomonic Projection

TL;DR

This paper introduces a CNN-based vanishing point detection method that uses an inverse gnomonic projection and synthetic data, achieving competitive results without scene assumptions.

Contribution

The novel approach employs a Gaussian sphere representation and synthetic training data, removing the need for labeled real images and scene assumptions.

Findings

Achieves competitive performance on horizon estimation datasets.

Utilizes synthetic data for training, reducing labeling effort.

Demonstrates versatility for additional applications.

Abstract

We present a novel approach for vanishing point detection from uncalibrated monocular images. In contrast to state-of-the-art, we make no a priori assumptions about the observed scene. Our method is based on a convolutional neural network (CNN) which does not use natural images, but a Gaussian sphere representation arising from an inverse gnomonic projection of lines detected in an image. This allows us to rely on synthetic data for training, eliminating the need for labelled images. Our method achieves competitive performance on three horizon estimation benchmark datasets. We further highlight some additional use cases for which our vanishing point detection algorithm can be used.