Giter VIP home page Giter VIP logo

Comments (6)

johannes-graeter avatar johannes-graeter commented on July 30, 2024

Hi there,

src/Image and tgt/Image are just intermediate topics of image preprocessing. The topic for the images is correct, however the feature matching takes way too long... it should be more in the range of 30 ms what is the resolution of your images. perhaps you could downscale them to kitti resolution? Did you compile it in release (O3)?

from limo.

johannes-graeter avatar johannes-graeter commented on July 30, 2024

Hi,

did you try with lower resolution, yet?

Best,

Johannes

from limo.

phonght32 avatar phonght32 commented on July 30, 2024

from limo.

johannes-graeter avatar johannes-graeter commented on July 30, 2024

I am honoured by people that invest time in trying to use my work and even file pull requests to my repo, so it is you that I must thank :)
Two cameras is not a problem, actually you only need one, either gray or color.
The 2d lidar scanner is a problem though: In the publication this is explained more in detail, but the base line is that at the current state you need lidar points around each successfull camera feature match. Then with an algorithm described in the paper the depth is extracted, mainly its a fancy form of interpolation on a plane. that this plane is stable however we need 3d points, with a line scanner no valid interpolation plane can be extracted.

Limo2 (branch on this repo) has also a nice monocular method, that estimates a groundplane for every keyframe. Merging it is one of my next todos :)

If you are motivated to develop for this, you could write a simple node that replaces mono_lidar_depth that assigns depth of feature matches to nearest points lidar points.
I tried in a 3h session, but on KITTI it didn't work well, perhaps for your setup it is better?

I am reworking this part however. My goal is to be able to use as less scan points as possible, however I am not sure if it will meet low computing power limitations since I plan to use dense optical flow from machine learning...

Best, Johannes

from limo.

johannes-graeter avatar johannes-graeter commented on July 30, 2024

Before you dive into that, I guess we should find out why the feature matching is so slow.
AFter merging limo2 I will isolate modules in order to run benchmarks with the raw kitti data. I think that would be a first step to see where to start tuning.

Best,
Johannes

from limo.

phonght32 avatar phonght32 commented on July 30, 2024

from limo.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.