/image-labeller

Image labeller for Salesforce Einstein Object Detection

Primary LanguageJavaScriptApache License 2.0Apache-2.0

Logo

Image Labeller for Salesforce Einstein Object Detection

This tool allows to label images in order to train a Salesforce Einstein Object Detection model.

Features:

  • Easy online installation via Heroku, local installation is also possible
  • Exports a zip file compatible with Salesforce Einstein Object Detection
  • Images hosted on Cloudinary
  • Responsive app build with Salesforce Lightning Design System
  • Proven to work on a project with 500 images and 700 tags

This tool is provided “as is“ without any warranty or support. Salesforce does not officially endorse it.

Editor screenshot Images screenshot
Labels screenshot Export screenshot

Installation

Heroku deploy (recommended)

Click on this button and follow the instructions to deploy the app:

Deploy

Setup screenshot

Local installation

This tool requires the following:

  • NPM
  • Node.js 8.7 or above
  • a PostgreSQL database
  • a Cloudinary account

Follow these steps to install the tool:

  1. Create a new PostgreSQL database and initialize it with the content of the db_init.sql file.

  2. Create a .env file with the following content and place it at the root of the project directory:

HTTPS=false
SESSION_SECRET_KEY='7#$rVXdTD86p'
DATABASE_URL='postgres://img:img@localhost:5432/img'
  1. Replace the values of these keys according to your settings:
Property Description
HTTPS Whether you use HTTPS to secure the connection to your server (generally false on a local setup)
SESSION_SECRET_KEY A random string used to secure your server sessions
DATABASE_URL URL to your PostgreSQL database

  1. Run the NPM installation script: npm install

  2. Run the program locally with either of these commands:

  • npm run build-n-start
  • heroku local (requires the Heroku CLI)