Giter VIP home page Giter VIP logo

Comments (8)

pengsongyou avatar pengsongyou commented on June 7, 2024

Ah, thanks for pointing this out. It may be because I was pre-processing the nuScenes dataset that has been preprocessed to the ply file by my colleague at Google. I will check on that. Otherwise, I think the scene.ply is simply the lidar clouds for every 0.5 seconds (which should be provided in the original dataset), and the GT semantic annotation was provided only on the last time stamp in each 0.5-second cloud.

You should be able to adapt the code without much change :)

Best
Songyou

from openscene.

kylegenova avatar kylegenova commented on June 7, 2024

Hi,

Sorry for the late reply. Here is an outline of how the scene.ply files were created for the nuScenes dataset:

For each keyframe in nuScenes

  1. Gather the preceding 0.5 seconds of LiDAR sweeps (everything up to the last keyframe, but not including it). Transform each sweep into the coordinate frame of the current keyframe.
  2. Create a mesh with one vertex per LiDAR point from all data in step 1 and no edges or faces.
  3. For each vertex, store 7 values- the x,y,z position (3 float32s), the r,g,b, values (3 Byte type values, using the RGB provided by nuScenes), the ground truth semantic category (one int32 value, according to the nuScenes lidarseg task category label).

I believe that is all the preprocessing that was done for the ply files- please let me know if anything is unclear, and sorry again for the delay (it's my fault, not Songyou's).

from openscene.

TyroneLi avatar TyroneLi commented on June 7, 2024

Could u release the processing code for nuscenes for this part? BTW, how to organize the file structure to obtain nuscenes_2d, including both the nuscens and nuscenes_image?

from openscene.

TyroneLi avatar TyroneLi commented on June 7, 2024

Ah, thanks for pointing this out. It may be because I was pre-processing the nuScenes dataset that has been preprocessed to the ply file by my colleague at Google. I will check on that. Otherwise, I think the scene.ply is simply the lidar clouds for every 0.5 seconds (which should be provided in the original dataset), and the GT semantic annotation was provided only on the last time stamp in each 0.5-second cloud.

You should be able to adapt the code without much change :)

Best Songyou

For Which paper we can refer to this release code? Could u give some more detailed cues?

from openscene.

pengsongyou avatar pengsongyou commented on June 7, 2024

Hi @TyroneLi

As I replied to you in the email, unfortunately we won't be able to release this part of the code since this preprocessing pipeline goes through internal google data formats, and it was not meant to be open sourced.

Hope you can still enjoy using our processed scene.ply files. Thanks!

Best
Songyou

from openscene.

TyroneLi avatar TyroneLi commented on June 7, 2024

Hi @TyroneLi

As I replied to you in the email, unfortunately we won't be able to release this part of the code since this preprocessing pipeline goes through internal google data formats, and it was not meant to be open sourced.

Hope you can still enjoy using our processed scene.ply files. Thanks!

Best Songyou

@pengsongyou Fine. But your processed nuscene_3d.zip only contained the val part.
https://github.com/pengsongyou/openscene/blob/main/scripts/download_dataset.sh
Could you upload the train set of your processed nuscenes_3d.
Thanks.

from openscene.

pengsongyou avatar pengsongyou commented on June 7, 2024

@TyroneLi thanks for reminding. I have already added the nuscenes_3d_train.zip to the download file. Please check it out.
https://github.com/pengsongyou/openscene/blob/main/scripts/download_dataset.sh

from openscene.

TyroneLi avatar TyroneLi commented on June 7, 2024

@TyroneLi thanks for reminding. I have already added the nuscenes_3d_train.zip to the download file. Please check it out. https://github.com/pengsongyou/openscene/blob/main/scripts/download_dataset.sh

Nice! Thanks!

from openscene.

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.