Data and codes for "SCANet: Self-paced semi-curricular attention network for non-homogeneous image dehazing"

<p dir="ltr">This record contains the data and codes for the paper "SCANet: Self-Paced Semi-Curricular Attention Network for Non-Homogeneous Image Dehazing" published in <i>2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)</i>. </p><h2>Requirement</h2><ul><li>Python 3.7</li><li>Pytorch 1.9.1</li></ul><h2><a href="https://github.com/gy65896/SCANet#network-architecture" target="_blank">Network Architecture</a></h2><h2><a href="https://github.com/gy65896/SCANet#train" target="_blank">Train</a></h2><ul><li>Place the training and test image pairs in the <code>data</code> folder.</li><li>Run <code>data/makedataset.py</code> to generate the <code>NH-Haze20-21-23.h5</code> file.</li><li>Run <code>train.py</code> to start training.</li></ul><h2><a href="https://github.com/gy65896/SCANet#test" target="_blank">Test</a></h2><ul><li>Place the pre-training weight in the <code>checkpoint</code> folder.</li><li>Place test hazy images in the <code>input</code> folder.</li><li>Modify the weight name in the <code>test.py</code>.<br></li></ul><pre><pre>parser.add_argument("--model_name", type=str, default='Gmodel_40', help='model name')<br></pre></pre><ul><li>Run <code>test.py</code></li><li>The results is saved in <code>output</code> folder.</li></ul><h2><a href="https://github.com/gy65896/SCANet#pre-training-weight-download" target="_blank">Pre-training Weight Download</a></h2><ul><li>The <a href="https://drive.google.com/file/d/15-M7bGwZkXtCato_kEfLi1VOq-tjblPL/view?usp=share_link" rel="nofollow" target="_blank">weight40</a> <code>Gmodel_40.tar</code> for the NTIRE2023 val/test datasets, i.e., the weight used in the NTIRE2023 challenge.</li><li>The <a href="https://drive.google.com/file/d/1ATye3j81n62VHXwGihShazYnMoEbTMLd/view?usp=share_link" rel="nofollow" target="_blank">weight105</a> <code>Gmodel_105.tar</code> for the NTIRE2020/2021/2023 datasets.</li><li>The <a href="https://drive.google.com/file/d/1sC81YfqOa82irk_Dy37I9oxpX4zniS2z/view?usp=share_link" rel="nofollow" target="_blank">weight120</a> <code>Gmodel_120.tar</code> for the NTIRE2020/2021/2023 datasets (Add the 15 tested images as the training dataset).</li></ul>