Skip to content

Latest commit

 

History

History
16 lines (10 loc) · 978 Bytes

README.md

File metadata and controls

16 lines (10 loc) · 978 Bytes

SimpleVO

A Python toy demo about visual odometry. I simply track features between current image and last keyframe and then perform triangulation. The demo does not have local BA optimization and is used to evaluate feature tracking/matching algorithms in the context of SLAM. The relative motion between current frame and last keyframe is estimated by OpenCV functions. This project is inspirsed and based on Python-VO and monocular_slam.

Note that the project is research code. The author is not responsible for any errors it may contain. Use it at your own risk!

Requirements

PyTorch, OpenCV, Pangolin, PyOpenGL ...

Usage

You need to download the outdoor pre-trained parameters of SuperGlue and save them in the /src/matcher/SuperGlue/weights/

python demo.py --config configs/params/kitti_**_**.yaml