Scenario Extraction from a Large Real-World Dataset

This repo contains a python implementation of scenario extraction from a large real-world dataset, i.e., Waymo Open Motion Dataset.

Dataset

Before running the code, please download the dataset v1.1 in tf_example format files.

Running the code

1. Install the packages in virtual environment with requirements.txt. The code is tested with Python 3.8.10.

2. To generate the tags, run with the following command.

   python .\utils\runner.py --data_dir your_data_path
   # An example of your_data_path is "./waymo_v_1_1/tf_example/training"

   The tags will be stored in an automatically created folder which is named by the time you run the code in ./results/

3. To categorize the scenarios by searching for a combination of tags, run the following command

   python .\utils\runner_sc.py --result_time your_result_time
   # MM-DD-HH_MM

Extracted scenarios

We have extracted a total of 215,090 scenarios across three scenario categories, which is accessible here.