yunnie121 / predrann

PredRANN

This is a Pytorch implementation of PredRANN, a method for radar echo sequence prediction (precipitation nowcasting) as described in the following paper:

PredRANN: The Spatiotemporal Attention Convolution Recurrent Neural Network for Precipitation Nowcasting, by Chuyao Luo, Xinyue Zhao, Xutao Li, Yunming Ye

Setup

Required python libraries: torch (>=1.4.0) + opencv + numpy + scipy (== 1.0.0) + jpype1. Tested in ubuntu + nvidia 3090Ti with cuda (>=11.0).

Datasets

We conduct experiments on CIKM AnalytiCup 2017 datasets: CIKM_AnalytiCup_Address or CIKM_Rardar

Training

To train the proposed model, it directly runs the CIKM_predrann.py

You might want to change the parameter and setting, you can change the details of variable ‘args’ in each files for each model

The preprocess method and data root path can be modified in the data/CIKM/data_iterator.py file

Prediction samples

5 frames are predicted given the last 10 frames.

Besides, we will offer some prediction results of models including PredRANN, PredRANN-L, and PredRANN-T as soon as possible.