Skip to content

Latest commit

 

History

History
24 lines (17 loc) · 1020 Bytes

README.md

File metadata and controls

24 lines (17 loc) · 1020 Bytes

Download dataset

Download the nuscenes dataset (the mini version is sufficient for now). This is required for rollout visualization, but not for training

  1. Go to https://www.nuscenes.org/login
  2. Register if don't already have an account
  3. Go to the download page and follow instructions

Set up conda environment

Navigate to the project root directory, In terminal, run

conda env create -f env.yml
conda activate nuscenes_env

To setup pythonpaths, in the project root directory, run source setup_pythonpath.sh local

Process Data

Dataset processing is divided into 2 stage

  1. The first stage extracts all relevant information from nuscenes and form a pandas dataframe. Run python process_data.py --mode=raw.
  2. The second stage takes the first dataframe as input and runs it through user defined filters to add high-level interaction labels / maneuvers. Run python process_data.py --mode=filter.

The filters functions reside in <project_root>/create_dataset/vehicle_behavior_filter/filters