The goal of this project is to upscale and improve the quality of low resolution images.
+- VRDL_HW4
+- models
+- checkpoint_srresnet.pth.tar
+- best_checkpoint_srresnet.pth.tar
+- dataset
+- testing_lr_images
+- testing_lr_images
+- 00.png
+- 01.png
...
+- 13.png
+- training_hr_images
+- training_hr_images
+- 2092.png
+- 8049.png
+- 8143.png
...
datasets.py
eval.py
model.py
split_train_val.py
test.py
train.py
utils.py
- Download the dataset: (I divide dataset into train and validation data)
https://drive.google.com/file/d/1n796E-LuV1lxqtXJvYgGQIKCWLpN5deH/view?usp=sharing
python train.py
Use the inference.ipynb on colab to generate the answer: https://colab.research.google.com/drive/1WwW7anJiJ9tDxJ93iKFvCDg9DRxixXYJ#scrollTo=4jL5aZH-2Jtu
You only need to cleck all the cells and you can get the output images in high-resolution image folder
My PSNR score: 27.4716 with 2100 epochs (training time 11hours 40mins)
https://blog.csdn.net/qianbin3200896/article/details/104181552