Parallel Interactive Transformer (PAIT)

This is the source code of the 7th place solution for stereo image super resolution task in 2022 CVPR NTIRE challenge (Team Name: No War).

Network Architecture:

An overview of our parallel interactive transformer network (The RDB and biPAM are the same as iPASSR).

Download the Results:

We share the quantitative and qualitative results achieved by our parallel interactive transformer on all the test sets for 4xSR. Results are available at Google Drive (including test images and our models).

Codes and Models:

Requirements:

PyTorch1.9.0，torchvision0.10.0. The code is tested with python=3.6, cuda=10.2. Matlab for prepare training data

Train:

• Run ./data/train/GenerateTrainingPatches.m to generate training patches.
• Run train_1 and _2.py to perform training. Checkpoint will be saved to ./log/

Test:

• Download the test sets and unzip them to ./data
• Run test_1 and _2.py to perform inference and calculate PSNR and SSIM scores.

Module Mean:

• Run mean_weights.py

Model Ensemble:

• Run ensemble_calculate.py

Challenge Result:

The official result of 2022 CVPR NTIRE challenge.

Acknowledgement:

• Thanks to the organizers of the 2022 CVPR NTIRE challenge.
• Thanks to my team members (Wenlong Guo and Zan Chen).

Any question regarding this work can be addressed to [email protected].