The Code for Traffic4cast 2022

Team: ustc-gobbler

Ranking: The first place

Download Links

• LONDON_2022.zip from HERE (2.8 GB)
• MADRID_2022.zip from HERE (4.0 GB)
• MELBOURNE_2022.zip from HERE (0.9 GB)
• T4C_INPUTS_2022.zip (1.0 GB)
• T4C_INPUTS_ETA_2022.zip (available September 2, 2022, 1.5MB)

After downloading and unzipping the data, please revise the data path in “t4c22_config.json”.

Prepare environment

conda env update -f environment.yml
conda activate t4c22

# Installing the torch geometric extras is optional, required only if using `torch_geometric`
# install-extras is not passed to pip from environment yaml, therefore add as post-step (https://github.com/conda/conda/issues/6805)
# replace with your CUDA version (cpu, ...), see https://pytorch-geometric.readthedocs.io/en/latest/notes/installation.html
CUDA="cu113"
python -m pip install -r install-extras-torch-geometric.txt -f https://data.pyg.org/whl/torch-1.11.0+${CUDA}.html

python t4c22/misc/check_torch_geometric_setup.py

Generate inputs and labels

Enter t4c22 folder and run the following commands.

python prepare_training_data_cc.py --data_folder [DATA_FOLDER]
python prepare_training_data_eta.py --data_folder [DATA_FOLDER]

Run models

You can choose to train the model from scratch, or use our trained ones for testing (put the save folder in the root).

train model for core challenge

python rec_cc.py --city [city] --device [gpu_id] --batch_size 2 --hidden_channels 32 --epochs 20 --fill -1

test model for core challenge

python rec_cc.py --city [city] --device [gpu_id] --batch_size 2 --hidden_channels 32 --epochs 20 --fill -1 --model_state test

train model for extended challenge

python rec_eta.py --city [city] --device [gpu_id] --batch_size 2 --hidden_channels 64 --epochs 50

test model for extended challenge

python rec_eta.py --city [city] --device [gpu_id] --batch_size 2 --hidden_channels 64 --epochs 50 --model_state test

Acknowledgements

This repository is based on NeurIPS 2022 Traffic4cast.