/DataTurks

ML data annotations made super easy for teams. Just upload data, add your team and build training/evaluation dataset in hours.

Primary LanguageJavaScriptApache License 2.0Apache-2.0

Features

NER tagging in Documents

Full length document annotations (PDF, Doc, Text etc).

Image Segmentation

    Draw free form polygons and generate image masks.

POS tagging

    A super easy interface to tag for PoS/NER in sentences.

DataTurks

[Note: As on March 2019, please use the branch master_with_dist as the main master branch has some build issues w.r.t dist folder]

Can run as a docker image as well. Here is the docker file specifying all the steps for setting things up:

https://github.com/DataTurks/DataTurks/blob/master/hope/docker/Dockerfile

If you rather have it run as a non-docker service, then see below.

Two main subcomponents:

  1. Hope: Java-mysql based backend.
Build:
Its a maven project, please install maven and then:
# cd hope
# mvn package -DskipTests <-- will build the .jar file.

Run:
The service is based on dropwizard and taken a config file on startup. This config file specifies the MYSQL end-points, 
password and the port to run the service on.

Setup mysql server as in: https://github.com/DataTurks/DataTurks/blob/master/hope/docker/mysqlInit.sql

# java -Djava.net.useSystemProxies=true -server -jar dataturks-1.0-SNAPSHOT.jar server onprem.yml
  1. Bazaar: React based front-end.

Mac Setup :

brew install node@8
brew link node@8
conda create -n bazaar python=2.7 anaconda
conda activate bazaar
xcode-select --install
sudo xcode-select -s /Applications/Xcode.app/Contents/Developer
rm -rf node_modules
npm rebuild node-sass
npm install
npm run dev  

Linux Setup: Install Node Js etc.

sudo apt-get -y install build-essential 
curl -sL https://deb.nodesource.com/setup_8.x | bash 	  
apt-get install --yes nodejs 	  
node -v 	  
npm -v  	  
npm i -g nodemon 	  
nodemon -v	  
apt-get clean 	  

Build:

  cd bazaar
  npm install && npm run build-onprem

Run the service:

  npm run start-onprem