# Unsupervised Monocular Depth and Ego-motion Learning with Structure and Semantics

## TL;DR

This paper introduces an unsupervised method for monocular depth and ego-motion estimation that incorporates structure and semantics, modeling object motion to improve accuracy in dynamic scenes.

## Contribution

It jointly learns 3D object motion, depth, and ego-motion using structure and semantics, advancing unsupervised monocular depth estimation in dynamic environments.

## Key findings

- Improved accuracy in dynamic scenes
- Joint modeling of object motion and depth
- Open-sourced code and models

## Abstract

We present an approach which takes advantage of both structure and semantics for unsupervised monocular learning of depth and ego-motion. More specifically, we model the motion of individual objects and learn their 3D motion vector jointly with depth and ego-motion. We obtain more accurate results, especially for challenging dynamic scenes not addressed by previous approaches. This is an extended version of Casser et al. [AAAI'19]. Code and models have been open sourced at https://sites.google.com/corp/view/struct2depth.