ngskcloud
/
label-3d-points
9
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
main
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from '78b80fa70f'
${ noResults }
label-3d-points/doc/install_from_source.md

1.9 KiB
Raw Blame History

Install

  1. clone the project

    git clone https://github.com/naurril/SUSTechPOINTS

  2. Install packages

    pip install -r requirement.txt

  3. Download model

    download pretrained model file deep_annotation_inference.h5, put it into ./algos/models

    wget https://github.com/naurril/SUSTechPOINTS/releases/download/0.1/deep_annotation_inference.h5  -P algos/models

Start

Run the following command in shell, then go to http://127.0.0.1:8081

python main.py

Object type configuration

Default object configuration is in obj_cfg.js

Adjust the contents to customize.

Data preparation

   +- data
       +- scene1
          +- lidar
               +- 0000.pcd
               +- 0001.pcd
          +- camera
               +- front
                    +- 0000.jpg
                    +- 0001.jpg
               +- left
                    +- ...
          +- aux_lidar
               +- front
                    +- 0000.pcd
                    +- 0001.pcd
          +- radar
               +- front_points
                    +- 0000.pcd
                    +- 0001.pcd
               +- front_tracks
                    +- ...
          +- calib
               +- camera
                    +- front.json
                    +- left.json
               +- radar
                    +- front_points.json
                    +- front_tracks.json
          +- label
               +- 0000.json
               +- 0001.json
       +- scene2

label is the directory to save the annotation result.

calib is the calibration matrix from point cloud to image. it's optional, but if provided, the box is projected on the image so as to assist the annotation.

check examples in ./data/example