Curb-GAN

The codes and data of paper "Curb-GAN: Conditional Urban Traffic Estimation through Spatio-Temporal Generative Adversarial Networks".

This is the overview of Curb-GAN. The implementation is realized by Pytorch.

Data Preparation

Since the whole dataset is huge, here we only provide sample datasets.

The sample data is located in data folder, please unzip data/speed_data.zip and data/inflow_data.zip and put the corresponding data files into traffic_speed_estimation/ and taxi_inflow_estimation/ folders, respectively.

Requirements for Reproducibility

• Cuda 9.2
• pytorch 0.4.1
• Python 3.6.7
• Devices: NVIDIA GTX 1080 GPUs
• System: Ubuntu 16.04

Training

Just take traffic speed estimation as an example:

1. cd traffic_speed_estimation/.
2. Running the codes: python train.py. (The parameters are defined in train.py)
3. The parameters of the trained model will be saved in to traffic_speed_estimation//CurbGAN_train.

The training process for taxi inflow estimation is similar.