STVDNet：Spatio Temporal Interactive Video Deraining Network

Requirements and Dependencies

• Ubuntu 22.04, cuda 11.3 or windows10 ,cuda 11.3
• Python 3.8, Pytorch 1.12.0, torchvision 0.13.0

Prepare Datasets

We provide datasets for testing, you can download NTURain datasets here. Password(sg4n).

Download pre-train model

We provide the pre-train model, you can download our checkpoint to test our network. pre-train. Password (kw7s)

Testing pipelines

You need firstly download the testing dataset of NTURain and decompression it.

• NTURain synthetic data set:

      python test.py --data_root /media/derain/NTU-derain --eval_file data/Dataset_Testing_Synthetic.json 
  

Results

Example results on the RainSynLight25 and NTURain. From left to right are Input, Efficient Derain, ESTINet, PReNet, VRGNet, STVDNet(ours).

Example result of real rain.