# Sparse2Dense: From direct sparse odometry to dense 3D reconstruction

## TL;DR

This paper introduces a deep learning-based monocular SLAM framework that constructs dense 3D models from sparse data using learned surface normals, improving accuracy in positioning and depth reconstruction.

## Contribution

It presents a novel approach combining sparse-to-dense mapping with learned surface normals and a single network for depth and normal prediction, advancing monocular dense SLAM.

## Key findings

- Significant improvement in visual tracking accuracy
- Enhanced depth prediction quality
- Outperforms state-of-the-art deep monocular dense SLAM methods

## Abstract

In this paper, we proposed a new deep learning based dense monocular SLAM method. Compared to existing methods, the proposed framework constructs a dense 3D model via a sparse to dense mapping using learned surface normals. With single view learned depth estimation as prior for monocular visual odometry, we obtain both accurate positioning and high quality depth reconstruction. The depth and normal are predicted by a single network trained in a tightly coupled manner.Experimental results show that our method significantly improves the performance of visual tracking and depth prediction in comparison to the state-of-the-art in deep monocular dense SLAM.