# image_navigation_python

**Repository Path**: Lee_Chao/image_navigation_python

## Basic Information

- **Project Name**: image_navigation_python
- **Description**: Python demo to our CVPR'19 publication: Mapping, Localization and Path Planning for Image-based Navigation using Visual Features and Map 
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2019-11-14
- **Last Updated**: 2020-12-17

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

 Python demo for our CVPR'2019 paper
# [Mapping, Localization and Path Planning for Image-based Navigation using Visual Features and Map](https://arxiv.org/pdf/1812.03795.pdf)

This code was tested with Python 3.5.5 and [mosek 8.1.80](https://www.mosek.com/downloads/).
The easiest way to install mosek is:
```
conda install -c mosek mosek
```

Please, execute 'demo.py' to view our demo.

This demo does the following using the concepts introduced in our paper:
1) Find landmarks in a subsequence of the Oxford Robotcar run from  2015-10-29 12:18:17
2) Match a short query sequence from 2014-11-18 13:20:12

The precalculated feature distances in this demo are based on features extracted with a _VGG-16 + NetVLAD + whitening_ network.
We use the _Off-the-shelf on Pitts30k_ model available on the [NetVLAD](https://www.di.ens.fr/willow/research/netvlad/) project page in combination with this [NetVLAD TensorFlow](https://github.com/uzh-rpg/netvlad_tf_open) implementation.   

If you do not have mosek installed, you can have a look at the saved
figures in the results folder instead.

The produced outputs are:
- Scatter plot of original reference and query sequences
- Topology of reference sequence used for finding landmarks with network flow
- Selected landmarks
- Accuracy vs. distance plot of the final matching