ml-inference-24

Docker file => Docker image => Docker container

1. Install Docker

• From this link, please install docker: https://www.docker.com/products/docker-desktop

  If the docker is already installed, you can skip this step.

2-1. Build Docker image manually

• Build the docker image from Docker file by running this command (root directory):

  docker build -t pytorch-docker .

• To check if the image has successfully built, please run this command:

  docker images

  You can check that docker image has built with the name pytorch-docker.

2-2. Use the pre-built Docker image

• If you want to skip manual docker image building process (2-1), please download the docker image in the root directory from this link: https://drive.google.com/file/d/10Z51z_KwiPyjVC1mf5b-9cIxJ7668ULW/view?usp=sharing

• Then, load docker image by running this command:

  docker load < pytorch-docker.tar.gz

  Windows: docker load -i pytorch-docker.tar.gz

• To check whether the image has successfully loaded, please run this command:

  docker images

3. Run the image as a container

• docker run -d -p 5000:5000 --name http-server pytorch-docker

  Docker runs in a detached mode (running in a background) if we use --detach or -d for short. Expose port 5000 inside the container to port 5000 outside the container, using the --publish flag on the docker run command. Name a container with http-server, by passing the --name flag to the docker run command.

• To check if the container is running, please run this command:

  docker ps

• In order to make sure that the server is running properly, please run this command:

  curl localhost:5000

  If the output is Server is Running!, it's running properly.

4. Query the server using images

• Please make sure that images are located in the root folder.

• Send the POST request with the test image file to the http server, by running the following curl command:

  curl -F "file=@[your test image file]" http://127.0.0.1:5000/upload

  (e.g, curl -F "[email protected]" http://127.0.0.1:5000/upload)

• Then, the server will take the image as an input, implement a simple image classification using a pre-trained Densenet-121 model for Pytorch, and return a classification as an output.

  (e.g, Imagenet_id : n01514668 , Classification : cock)

• After the testing is finished, please run this command and stop the container running:

  docker stop http-server

5. Test script

• To test this server with test images provided in this folder, please run test.sh script.

• Before you run this test script, please make sure Docker is running on your computer.

• Then, run this command in the root folder.

  bash test.sh