# Tensorflow 2.8.0 + Pytorch 1.10.2
- 開發環境 window 10
- CPU: i7-9700F
- GPU: RTX2070 GDDR6GB
- Memory: DDR4 20GB
- IDE: Jupyter Notebook
- python 3.9
- tensorflow-gpu 2.8.0

    pip install tensorflow-gpu==2.8.0

- torch 1.10.2+cu113
- https://anaconda.org/pytorch/pytorch
- cudatoolkit 11.2.0
- https://anaconda.org/esri/cudatoolkit
- cudnn 1.5.1
- https://anaconda.org/esri/cudnn
- 測試安裝
- Tensorflow

        python
        import tensorflow
        print(tensorflow.config.list_physical_devices('GPU'))
        #from tensorflow.python.client import device_lib
        #print(device_lib.list_local_devices())

- Pytorch

        python
        import torch
        print(torch.cuda.is_available())

- 其他參考套件
- kornia==0.6.3
- opencv-python-headless==4.5.5.62
- scikit-learn==1.0.2
- Pillow==9.0.1
- pandas==1.4.1
- openpyxl==3.0.9
- Flask==2.0.3
- yacs==0.1.8
- pylint==2.12.2
- black==22.1.0