Facial landmark annotation tool(FLAT)

Realse Note

V0.1.8-beta

• add Dcokerfile.
• upgrade landmark model file.

V0.1.7

Update date 2020/3/19

• fix for window

install

• Ubuntu

conda create -n flat python==3.7
conda activate flat
# install protobuf
pip install protobuf==3.19
# install tensorflow
conda install tensorflow==1.14
or 
pip install tensorflow==2.3
# install pyqt5
conda install pyqt==5.12.3
or 
pip install PyQt5==5.12.3
# install opencv
conda install opencv==4.1.0
or 
pip install opencv-python==4.1.0.25

• window10 isntall

1. 下载并安装anaconda 下载地址： https://www.anaconda.com/distribution/

2. 创建conda环境 添加清华源,打开"Anaconda Powershell prompt"输入如下命令：

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --append channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/fastai/
conda config --append channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
conda config --append channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/bioconda/
conda config --set show_channel_urls yes

2. 安装软件

• 安装pip 参考地址：https://blog.csdn.net/esting_tang/article/details/80973478

• 安装tensorflow

conda create -n flat python=3.7
conda activate flat
conda install tensorflow==1.14.0

• 安装OpenCV pip install opencv-python==4.1.0.25
• 安装pyqt conda install pyqt==5.12.3

运行

python main.py

pyinstaller打包应用程序

pyinstaller -F main.py -n flat_v0.1.3_for_ubuntu18.04

• pyinstaller可能遇到的问题

+ install pyinstaller
pip install pyinstaller
# pyinstaller package run error
"No module named 'pkg_resources.py2_warn'"
setuptools降级到44.0.0

conda install setuptools==44.0.0

+ pyinstaller cmd
pyinstaller -F main.py -n flat_v0.1.6_for_win10

docker

build

docker build --tag=ts/flat:0.1.7 .

create

sudo apt-get install x11-xserver-utils
export DISPLAY=:0
xhost +
docker create --name flat -it -e DISPLAY=$DISPLAY -p 8888:8888 -v /tmp/.X11-unix:/tmp/.X11-unix -v ~/iwork/share:/mnt:rw ts/flat:0.1.7

start & run

1. docker ps -a 查看container

$ docker ps -a
CONTAINER ID        IMAGE                              COMMAND             CREATED             STATUS                         PORTS               NAMES
5a22b5bd015a        ts/flat:0.1.7                      "/bin/bash"         5 seconds ago       Created                                            flat

2. run

docker start 5a22b5bd015a
docker exec -it 5a22b5bd015a /bin/bash
root@5a22b5bd015a:/flat#

ui to py

pyuic5 -o output.py test.ui