Skip to content

Build and deploy YMIR on mac laptop (M1 chip)

Jump to bottom
Phoenix edited this page Nov 17, 2022 · 7 revisions

Deploy YMIR on MacOS platform (Apple M1 chip)

Last updated: 11/17/2022, revised for Release 2.0.0

Current deploy script ymir.sh only supports Linux platform. For developers that hopes to try YMIR on their own MacOS machines, there are some customized steps:

  1. build YMIR backend/web docker image
  2. modify .env file
  3. modify docker-compose.yml

Build YMIR docker image

  1. Add gcc to ymir/Dockerfile.backend, as the base docker image does not include gcc environment, which are required for some python perquisites.

  2. Build backend:

docker build -t industryessentials/ymir-backend:release-2.0.0 --build-arg PIP_SOURCE=https://pypi.mirrors.ustc.edu.cn/simple --build-arg GO_SOURCE='https://mirrors.aliyun.com/goproxy/,direct' -f Dockerfile.backend .

  1. Build web docker image:

docker build -t industryessentials/ymir-web:release-2.0.0 --build-arg NPM_REGISTRY='https://registry.npm.taobao.org' .

Manually start docker-compose

  1. modify .env file

add IP address, set ANONYMOUS feedback, change SERVER_RUNTIME to runc.

  1. modify docker-compose.yml

use volnet/tensorflow-tensorboard to replace tensorflow/tensorflow:2.9.1

Optional: start label tool

  1. label-studio supports MacOS, so no change is needed.
  2. for label-free, which uses mysql:5.7.22, an alternative option is mariadb:10.5.8

That's it, enjoy your journey!

docker-compose -f docker-compose.yml down && docker-compose -f docker-compose.yml up -d

Clone this wiki locally