Learning Depth from Monocular Videos using Direct Methods

TL;DR

This paper introduces a novel unsupervised method for learning depth from monocular videos by integrating direct visual odometry and a new depth normalization technique, eliminating the need for a pose predictor.

Contribution

It proposes a direct visual odometry-based approach to learn depth without a separate pose CNN, improving performance over existing monocular video training methods.

Findings

Outperforms state-of-the-art methods in depth prediction accuracy.

Incorporating differentiable DVO enhances learning stability.

Depth normalization significantly improves results.

Abstract

The ability to predict depth from a single image - using recent advances in CNNs - is of increasing interest to the vision community. Unsupervised strategies to learning are particularly appealing as they can utilize much larger and varied monocular video datasets during learning without the need for ground truth depth or stereo. In previous works, separate pose and depth CNN predictors had to be determined such that their joint outputs minimized the photometric error. Inspired by recent advances in direct visual odometry (DVO), we argue that the depth CNN predictor can be learned without a pose CNN predictor. Further, we demonstrate empirically that incorporation of a differentiable implementation of DVO, along with a novel depth normalization strategy - substantially improves performance over state of the art that use monocular videos for training.