codes/README_with_dev.md

##1.提交镜像并push到服务器上

sdk

原始镜像： face2face_train:0.0.1_origin_sdk（12.1G）
镜像保存为tar文件： docker save -o face2faceweb.tar bedcce862ec3
从当前容器生成镜像：
docker commit -m="facesdk init" -a="jc"  face2facesdk:
将新镜像推到docker服务器上：
# 1. 先把本地镜像名称和tag改成公司仓库的名称
docker tag face2facesdk: base.image.docker:5000/python/face2face_train:_sdk
# 2. 推送修改后的镜像到公司仓库
docker push base.image.docker:5000/python/face2face_train:_sdk

base.image.docker:5000/python/face2face_train:0.0.3_sdk  13.1G
python app_local.py

sdk slim

从源镜像到slim: 1.pip install Cython -i https://pypi.tuna.tsinghua.edu.cn/simple 2.执行package: https://digital-public.obs.cn-east-3.myhuaweicloud.com/vpp/1batchSynth/2/packaging_script.py 3. 删除build下面的临时so包

 face2facesdk_slim:
docker tag face2facesdk_slim: base.image.docker:5000/python/face2face_train:_sdk_slim
docker push base.image.docker:5000/python/face2face_train:_sdk_slim

web

docker commit -m="faceweb init" -a="jc"  face2faceweb:
docker tag face2faceweb: base.image.docker:5000/python/face2face_train:_web
docker push base.image.docker:5000/python/face2face_train:_web   550M
python -m gradio_web.gradio_test

2.依赖项

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
pip install   opencv-python-headless opencv-contrib-python-headless librosa onnxruntime-gpu einops cv2box -i https://mirrors.aliyun.com/pypi/simple
pip install  scikit-image  matplotlib pyyaml apstone  numexpr flask -i https://mirrors.aliyun.com/pypi/simple
pip install typeguard==2.13.3 -i https://mirrors.aliyun.com/pypi/simple

365 248 275 7h4no3   GJ20230901
guiji123
180.96.14.117 base.image.docker
docker login  base.image.docker:5000
admin/Guiji@123!