Solution for our Neurips 2019 challenge on traffic movie forecasting - top-6 teams.
The high-resolution traffic map videos for the challenge were derived from positions reported by a large fleet of probe vehicles over 12 months, and are based on over 100 billion probe points. Each video frame summarizes GPS trajectories (of 3 channels, encode speed, volume, and direction of traffic) mapped to spatio-temporal cells, sampled every 5 minutes and of size 495 ˆ 436.
- Tensorflow version >= 1.14.1
- Python version >= 3.6
- For training new models, you'll need an NVIDIA GPU
trainer.pyfor training the model
Apache 2.0