5.9 KB


Synthesis Image Noise Detection on Distributed Data (SIN3D)

A web app to collect data on noise detection by humans on images.

Download project

git clone
cd SIN3D



Developer documentation


API documentation


Data extraction

You can use the SIN3D-data-extract library to easily extract data from this application database.

Run as a Docker instance

Configure application

Use the following environment variables to configure the application.

Option Default value Description Server Client
PORT 5000 The port used by the started application
SERVE_CLIENT true Should the server serve client (Fully local application) ⬜️
IMAGES_PATH ./images The directory where the images are stored (absolute path if changed ⚠️) ⬜️
MONGO_URI mongodb://localhost/sin3d MongoDB database connection URI ⬜️

Configure more deeply the way the app works by modifying config.js.

Option Default value Description
apiPrefix /api The url prefix for the API
imageServedUrl /api/images The url prefix from where the images are served
serverPort 5000 The port used by the server
mongoDatabaseURI mongodb://localhost/sin3d MongoDB database connection URI
imagesPath ./images The directory where the images are stored
serveClient true Should the server serve client files from the /dist directory
fileNameConvention /^(.*)?_([0-9]{2,})\.(.*)$/ File name convention for images
extractsDirName extracts Name of the directory containing extracts
sceneFileNameBlackList ['config', 'seuilExpe', 'extracts'] Files to ignore in scenes
deleteExtractsCronTime 0 3 * * * (every day at 03:00 AM) Cron time for extracts deletion
logger Logs : logs/server.combined.log Errors : logs/server.error.log Default application logger
dbLogger Logs : logs/db.log Errors : logs/db.error.log Database logger configuration

Run server + client


PORT=8080 SERVE_CLIENT=true IMAGE_PATH=/var/images docker-compose up


SET IMAGE_PATH=C:\my\images
docker-compose up

Run server only


SERVE_CLIENT=false docker-compose up


docker-compose up

Run client only

docker-compose -f docker-compose.frontapp_only.yml up


New version Docker build

When using a new version of the project, you need to re-build the application with Docker.

# Server / Server + client versions
docker-compose build

# Client only version
docker-compose -f docker-compose.frontapp_only.yml build

Docker instance on a Windows host

As of now, Windows hosts are not supported due to MongoDB volumes not being mappable to this system.

Run on the file system

Project setup

Install project's dependencies.

yarn install

If this is a production build, set the NODE_ENV environment variable to production to enhance performance.


export NODE_ENV=production


SET NODE_ENV=production


Run the server

yarn run server:start

If you specifically want to not start the extracts remover service (extracts are automatically deleted everyday), use the following.


Automatically fix the API code syntax with ESLint

yarn run server:lint


To use the client, you need to initialize experiment configurations first.

cp -r experimentConfig.default experimentConfig

You can modify the configuration for each mixins, experiments or scene by modifying files in the newly created ./experimentConfig directory.

Compile and minify for production

Files will be built to the dist/ directory.

yarn run app:build

Run a hot-reload server for client development

yarn run app:dev

Automatically fix the client code syntax with ESLint

yarn run app:lint

Automated deployment

The app can be automatically deployed when a push event is sent to the master branch using Gogs. Open a port to the web and start webhook_deploy_gogs.js.

WEBHOOK_SECRET=your_webhook_secret WEBHOOK_PORT=5000 SERVE_CLIENT=true PORT=8080 node webhook_deploy_gogs.js

You can pass any parameters to the script, they will be passed to the Docker instance. The following are required.

Option Description
WEBHOOK_SECRET The secret set on Gogs to verify the identity
WEBHOOK_PORT The port the script is listening to

Running tests

The docker-compose script will automatically run tests. Use the following command to run them by hand.

yarn test

Generate API documentation

The docker-compose script will automatically build the API documentation. Use the following command to build it by hand.

yarn doc

The documentation is generated to the doc/ directory. If you started the server, the documentation is served at /doc/.

Collected DATA

During application use a lot of data is collected by platform, here a description of main message identifier used:

  • COLLECT_DATA: specify that the collected data process began.
  • EXPERIMENT_STARTED: message which is used for start of an experiment.
  • EXPERIMENT_DATA: message for data of an experiment.
  • EXPERIMENT_VALIDATED: end of experiences message.


The MIT license