Skip to content

Upload via links plugin

If your images are storing on the server that available from the browser you can use "Links" plugin, to use them on the Supervisely without duplicating.

http://192.168.1.49:4321/dataset_1/00003.png
http://192.168.1.49:4321/dataset_100004.png

Each link should be placed on a separate line. After importing into Supervisely, a separate dataset will be created for each .txt file.

Step 2: Go to the Explore page

Select "Plugins" tab. Find "Links" plugin in the list and click "Add" button.

Step 3: Import files

Go to the "Import" page. Select "Links" plugin and drag and drop your .txt files.

Enter project name and click "Start import" button.

Import will be started after files uploaded to the server.

After task completed go to the "Projects" page

Links generation

For the following steps you need have docker and docker-compose to be installed

Create 'docker-compose.yml' with following content

version: '2.2'

services:
  app:
    image: supervisely/links-plugin-txt-generator
    environment:
      - IMAGES_SERVER_PATH=<images_server_url>
    volumes:
    - <path_to_your_images_folder>:/tmp/data-src:ro
    - <path_to_out_txt_files>/:/tmp/data-dest

If you doesn't have images server, you can add following code in the docker-compose

Setup images server

port must be available from the browser

  server:
    image: nginx:1.15.8-alpine
    volumes:
    - <path_to_your_images_folder>:/usr/share/nginx/html:ro
    ports:
    - "<port>:80"

For example:

version: '2.2'

services:
  app:
    image: supervisely/links-plugin-txt-generator
    environment:
      - IMAGES_SERVER_PATH=http://192.168.1.49:4321
    volumes:
    - ./data/:/tmp/data-src:ro
    - ./output/:/tmp/data-dest

  server:
    image: nginx:1.15.8-alpine
    volumes:
    - ./data/:/usr/share/nginx/html:ro
    ports:
    - "4321:80"

Next run the following command in the terminal:

  docker-compose up -d server
  docker-compose up app

After app container complete execution, you can import files from <path_to_out_txt_files> folder as described above.