Skip to content
Andrew Edgley edited this page Feb 22, 2023 · 1 revision

Additional dependencies

This script relies on the tensorflow and waymo-open-dataset-tf-2-4-0 packages.

Note: waymo-open-dataset-tf-2-4-0 is not available on Windows You can augment your pipenv by doing the following. Commands should be similar in Windows, ensure that you are in the src folder.

#Move to src directory cd src #Install extra dependencies pipenv install tensorflow pipenv install waymo-open-dataset-tf-2-4-0 

Running the conversion tool

pipenv run python3 waymo-ct.py -f <input_path> -o <output_path> -r <set recursive flag> 

The conversion tool converts nuScenes data to the LVT generic format data for use in the visualization tool.

Launch Options

  • -f Specifies the input file. This should be a TFRecord File
  • -o The output path and name of the LVT directory that will be created
  • -r Specifies that the input file is a directory containing only TFRecord files. The tool will then convert all of the TFRecord files inside that directory. Used for batch processing

Predicted Data

The waymo conversion tool does not currently implement our function to import predicted data Running the export tool The export tool converts LVT generic format data back to a nuScenes dataset.

pipenv run python3 waymo-rt.py -f <input_path> -o <output_path> -r <set recursive flag> 

Clone this wiki locally