# OpenCV

## 環境
### AnaConda with Sublime Text

> [如何优雅的融合Anaconda和Sublime Text3?](https://zhuanlan.zhihu.com/p/289860842) `下方留言`
> [解決Sublime text3關掉白色邊框](https://www.itread01.com/content/1548698600.html)

### CUDA 不相容 (torch.cuda.is_available()=False)
:::success
解決方案：檢查顯卡支援的版本後，查看對應的Pytorch及相關套件版本，在Anaconda裡配置Environment，再修改Sublime Text裡的Build指令
`activate CVDL`
:::

conda create --name CVDL python=3.6
activate CVDL

conda install pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cudatoolkit=11.0 -c pytorch
pip install torchsummary tqdm matplotlib

## OCW
- NTU ML
- [Youtube](https://www.youtube.com/watch?v=bHcJCp2Fyxs&list=PLJV_el3uVTsMhtt7_Y6sgTHGHp1Vb2P2J)
- [Course Website](https://speech.ee.ntu.edu.tw/~hylee/ml/2021-spring.html)

## Ref
- PyQt5
- https://www.wongwonggoods.com/python/pyqt5-1/

- Q1.4 Trackbar
- https://blog.csdn.net/weixin_44493841/article/details/102483184
- http://www.1zlab.com/wiki/python-opencv-tutorial/opencv-trackbar/

- Q3 Padding
- [OpenCV-Python: cv2.copyMakeBorder()函数详解](https://blog.csdn.net/qq_36560894/article/details/105416273)
- [opencv学习——padding扩充边界](https://blog.csdn.net/ei1990/article/details/78350974)
- Q5 Progress Bar
- [[Python] 使用 tqdm 套件展示進度條、並嘗試自己寫進度條](https://clay-atlas.com/blog/2019/11/11/python-chinese-tutorial-tqdm-progress-and-ourself/)

- train
- [PPT範例](https://www.pyimagesearch.com/2021/07/19/pytorch-training-your-first-convolutional-neural-network-cnn/)
- [PyTorch - 練習kaggle - Dogs vs. Cats - 使用 Transfer Learning - Fine tune 預訓練模型](https://hackmd.io/@lido2370/HyLTOlSn4?type=view)
- 只有改FC層的輸出(1000->2)
- https://clay-atlas.com/blog/2019/10/20/pytorch-chinese-tutorial-classifier-cifar-10/
- https://xmfbit.github.io/2017/02/25/pytorch-tutor-01/
- https://blog.csdn.net/qq_29367075/article/details/112760843
- [CNN02：Pytorch实现VGG16的CIFAR10分类](https://blog.csdn.net/xiaoheizi_du/article/details/89365916)
`最後一段`