Vision-Kickstarter

Docker based ML / CV project skeleton

Create Development Environment in PyCharm

Clone new project and create run/debug configurations with below settings in PyCharm

Add Docker Server

Open Preferences > Build,Execution,Deployment > Docker : Add docker server

Create Dockerfile Configuration and Build Image

Create new run/debug configuration from Dockerfile template with below settings: (You should change image tag, container name and project root directory name according to repo name)

• Image Tag: vision-kickstarter
• Container name: vision-kickstarter
• Context folder: .
• Bind ports: 8888:8888
• Bind mounts: {project-root-path}:/opt/project
• Run build image: checked

Run this file to build image

Creating Container From Image

Create new run/debug configuration Docker Image with below settings:

• Name : docker container
• Server :Docker
• Image ID or name: vision-kickstarter
• Contanier name : vision-kickstarter
• Bind ports: 8888:8888
• Bind mounts: {project-root-path}:/opt/project

Run this file to build container

Create Docker Remote Interpreter

Image building should be finished for create docker remote interpreter.
Open Preferences > Project > Python interpreter > Add Python Interpreter : Select "Docker" and set "Image name" as "vision-kickstarter" then click "Ok"
Configure "Path mappings" setting in Python Interpreter: Open "Edit Project Path Mappings" dialog window
Add new Path Mappings with below settings:

• Local path: {project-root-path}
• Remote path: /opt/project

Create Flask Server Configuration and Start (api.py)

Create new run/debug configuration from "Flask server" template with below settings: (You should change Python Interpreter according to image tag)

• Target type: Script path
• Target: {project-root-path}/apps/api.py
• Additional Options: --host=0.0.0.0 --port=5000
• FLASK_ENV: development
• FLASK_DEBUG: checked
• Python Interpreter: vision-kickstarter:latest
• Docker container settings
  • Port bindings:
    • Host Port: 5000
    • Container Port: 5000
    • Host IP: 0.0.0.0
    • Protocol: tcp
    • run options : --entrypoint= --rm --name=vision-kickstarter-api
  • Volume bindings: {project-root-path}:/opt/project

Run this file to start flask server on container

Run Python App on Running Container in PyCharm

Right click on running container at "Services" window and then select option to "Create terminal" or "Exec"
Write that command template for run python app in exec or terminal: "python {path-of-app-file}"

• Run inference.py: python apps/inference.py

Create Development Environment in Terminal

Clone new project and create new image and container in project root path

Build Image

docker build -t vision-kickstarter .

Create Container

docker run -v {project-root-path}:/opt/project -p 8888:8888 --name vision-kickstarter -it vision-kickstarter

Access Jupyter Notebook Server Token and Links

docker exec vision-kickstarter jupyter notebook list

Run Python App in Container

docker exec {container-name} python {path-of-app-file}

Run inference.py app on Flask (apps/api.py) container:

docker exec {flask-service-container-id} python apps/inference.py

Run train.py app on vision-kickstarter container:

docker exec vision-kickstarter python apps/train.py