Camera-based vehicle velocity estimation from monocular video

TL;DR

This paper presents a real-time, lightweight monocular video-based vehicle velocity estimation method that outperforms competing approaches and is suitable for autonomous driving applications.

Contribution

It introduces a novel, simple regression approach using trajectory features and provides an extensive study of feature effectiveness for monocular velocity estimation.

Findings

Outperforms all challenge entries in velocity estimation accuracy

Light-weight trajectory features outperform deep learning-based depth and motion cues

Achieves real-time performance on a single CPU

Abstract

This paper documents the winning entry at the CVPR2017 vehicle velocity estimation challenge. Velocity estimation is an emerging task in autonomous driving which has not yet been thoroughly explored. The goal is to estimate the relative velocity of a specific vehicle from a sequence of images. In this paper, we present a light-weight approach for directly regressing vehicle velocities from their trajectories using a multilayer perceptron. Another contribution is an explorative study of features for monocular vehicle velocity estimation. We find that light-weight trajectory based features outperform depth and motion cues extracted from deep ConvNets, especially for far-distance predictions where current disparity and optical flow estimators are challenged significantly. Our light-weight approach is real-time capable on a single CPU and outperforms all competing entries in the velocity estimation challenge. On the test set, we report an average error of 1.12 m/s which is comparable to a (ground-truth) system that combines LiDAR and radar techniques to achieve an error of around 0.71 m/s.