{%hackmd @expectlai/darktheme %}
# 資專研究札記By賴昱錡 (Nov.&Dec.)
## 2022/11/8
Comparison of deep learning methods for colorizing historical aerial imagery
https://meetingorganizer.copernicus.org/EGU22/EGU22-7686.html

ColorAI -Automatic Image Colorization using CycleGAN
https://www.researchgate.net/publication/350957456_ColorAI_-Automatic_Image_Colorization_using_CycleGAN

CycleGAN model:
https://github.com/junyanz/CycleGAN
https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix
+ Create a dataset folder under /dataset for your dataset.
+ Create subfolders testA, testB, trainA, and trainB under your dataset's folder. Place any images you want to transform from a to b (cat2dog) in the testA folder, images you want to transform from b to a (dog2cat) in the testB folder, and do the same for the trainA and trainB folders.

Tensorflow Datasets:
https://www.tensorflow.org/datasets?hl=zh-tw

## 2022/11/10

Visdom's Github
https://github.com/fossasia/visdom

CycleGan-horse2zebra
https://machinelearningmastery.com/cyclegan-tutorial-with-keras/

GANILLA:
https://github.com/giddyyupp/ganilla
https://arxiv.org/abs/2002.05638

## 2022/11/15

各種 GAN 的延伸:
https://github.com/NVIDIA/pix2pixHD
https://github.com/taesungp/contrastive-unpaired-translation

https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix/blob/master/docs/tips.md#notes-on-colorization

https://yanwei-liu.medium.com/wandb-weights-and-biases-a-better-choice-over-tensorboard-6c75b5f972b1

## 2022/11/17

pix2pix原理
https://blog.csdn.net/weixin_36474809/article/details/89004841

CycleGAN原理
https://blog.csdn.net/qq_40520596/article/details/104714762

Python 的 gdown 套件
https://clay-atlas.com/blog/2020/03/13/python-chinese-note-package-gdown-download-google-drive/

各種 GAN
https://github.com/eriklindernoren/PyTorch-GAN

2022/11/25

pix2pix model 的 instruction
`
Change the --dataroot and --name to your own dataset's path and model's name. Use --gpu_ids 0,1,.. to train on multiple GPUs and --batch_size to change the batch size. Add --direction BtoA if you want to train a model to transfrom from class B to A.
`

maptiles 的截圖
https://github.com/riatelab/maptiles/
https://github.com/AliFlux/MapTilesDownloader

Linux scp指令應用:
https://blog.gtwang.org/linux/linux-scp-command-tutorial-examples/

Kaggle pix2pix datasets
https://www.kaggle.com/datasets/vikramtiwari/pix2pix-dataset

## 2022/12/1
pandas&matplotlib
https://www.learncodewithmike.com/2021/03/pandas-and-matplotlib.html

## 2022/12/5
Notes on Colorization

No need to run combine_A_and_B.py for colorization. Instead, you need to prepare natural images and set --dataset_mode colorization and --model colorization in the script. The program will automatically convert each RGB image into Lab color space, and create L -> ab image pair during the training. Also set --input_nc 1 and --output_nc 2. The training and test directory should be organized as /your/data/train and your/data/test. See example scripts scripts/train_colorization.sh and scripts/test_colorization for more details.

## 2022/12/8

https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix/issues/1161

https://paperswithcode.com/task/image-colorization?fbclid=IwAR0AI_17vmTLvfdrpL_chgWyOcGFon-Q7aa54IxFh1cXxUJ2tJ6pKu74EpM

## 2022/12/14
找度量模型成效的方法:

FID score
https://github.com/mseitzer/pytorch-fid
https://blog.csdn.net/weixin_43135178/article/details/120621393

## 2022/12/15

FID evaluating
pytorch_fid ./Fake ./Real
FID: 49.12806924497005

pix2pix testing options

def initialize(self, parser):
        parser = BaseOptions.initialize(self, parser)  # define shared options
        parser.add_argument('--results_dir', type=str, default='./results/', help='saves results here.')
        parser.add_argument('--aspect_ratio', type=float, default=1.0, help='aspect ratio of result images')
        parser.add_argument('--phase', type=str, default='test', help='train, val, test, etc')
        # Dropout and Batchnorm has different behavioir during training and test.
        parser.add_argument('--eval', action='store_true', help='use eval mode during test time.')
        parser.add_argument('--num_test', type=int, default=50, help='how many test images to run')
        # rewrite devalue values
        parser.set_defaults(model='test')
        # To avoid cropping, the load_size should be the same as crop_size
        parser.set_defaults(load_size=parser.get_default('crop_size'))
        self.isTrain = False
        return parser

CUT model
https://blog.csdn.net/qq_37614597/article/details/114671050

科展報告書撰寫模板
https://twsf.ntsec.gov.tw/activity/race-1/58/pdf/NPHSF2018-052511.pdf

初步測試結果:
https://hackmd.io/@expectlai/rkkJffR_s

TTUR
https://github.com/bioinf-jku/TTUR