Repository for "Learning Parallax Attention for Stereo Image Super-Resolution", CVPR 2019
Pytorch implementation of "Learning Parallax Attention for Stereo Image Super-Resolution", CVPR 2019
Figure 1. Overview of our PASSRnet network.
Figure 2. Illustration of our parallax-attention mechanism.
Figure 3. A toy example illustration of the parallax-attention and cycle-attention maps generated by our PAM. The attention maps (30×30) correspond to the regions (1×30) marked by a yellow stroke. In (a) and (b), the first row represents left/right stereo images, the second row stands for parallax-attention maps, and the last row represents cycle-attention maps.
Figure 4. The Flickr1024 dataset.
python=3.6, cuda=9.0)
data/train/Flickr1024(Note: In our paper, we also use 60 images in the Middlebury dataset as the training set.)
data/trainand run
generate_trainset.mto generate training data.
python train.py --scale_factor 4 --device cuda:0 --batch_size 32 --n_epochs 80 --n_steps 30
testing/colored_0and
testing/colored_1in
data/test/KITTI2012/original
data/testand run
generate_testset.mto generate test data.
data/testas below:
data └── test ├── dataset_1 ├── hr ├── scene_1 ├── hr0.png └── hr1.png ├── ... └── scene_M └── lr_x4 ├── scene_1 ├── lr0.png └── lr1.png ├── ... └── scene_M ├── ... └── dataset_N
python demo_test.py --scale_factor 4 --device cuda:0 --dataset KITTI2012
Figure 5. Visual comparison for 2× SR. These results are achieved on “testimage013” of the KITTI 2012 dataset and “testimage019” of the KITTI 2015 dataset.
Figure 6. Visual comparison for 4× SR. These results are achieved on “testimage004” of the KITTI 2015 dataset.
Figure 7. Visual comparison for 2× SR. These results are achieved on a stereo image pair acquired in our laboratory.
@InProceedings{Wang2019Learning, author = {Longguang Wang and Yingqian Wang and Zhengfa Liang and Zaiping Lin and Jungang Yang and Wei An and Yulan Guo}, title = {Learning Parallax Attention for Stereo Image Super-Resolution}, booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)}, year = {2019}, }
For questions, please send an email to [email protected]