📷 PictoPercept

PictoPercept is a fully open-source survey toolkit designed to study human and machine preferences in an increasingly digitized world. PictoPercept provides visual choice scenarios, where one can make choices by comparing pairs of facial images in short time frames. You can try out sample surveys here.

🌱 Supported By

• Amsterdam School of Communication Research, University of Amsterdam
• Digital Communication Methods Lab, University of Amsterdam

Reach out to saurabh.khanna@uva.nl with feedback and/or questions.

Project structure

• pictopercept -> Holds the web/Flask's python backend code, such as the views.py and db.py, that define the routes and what gets shown.
• pictopercept/templates -> Where all the HTML files are located, which are the final web pages that the user will see. They use basic Flask/Jinja2 templating, to pass variables from Python to them.
• pictopercept/static -> Holds the static files such as JavaScript and CSS code, fonts and images.
• pictopercept/lib -> A custom library that defines custom types and default survey handlers to make defining new surveys easier.
• pictopercept/surveys -> Here is where you should place all the custom surveys to show in the website. You can view the base survey definition at pictopercept/lib/common_types.py to view all the needed attributes and functions to implement per survey.

Developing

First, make sure to check Environment Variables and Storage and Database to setup everything properly.

It is recommended to use some sort of virtual environment, such as venv (there are other alternatives such as pipenv or conda), which will help get and manage the correct versions of python-related libraries:

python3 -m venv .venv

Then, to activate the environment:

source .venv/bin/activate

Once done, you can execute the following to install all the python dependencies:

pip3 install -r requirements.txt

Then, run the server using:

flask run --debug

This command will start the local website in the port 5000. You can visit http://localhost:5000 in your browser to view it.

Environment Variables

There is a provided .env.example where you can view all the env variables used by our program.

These are:

• FLASK_SECRET_KEY: A random hex key that Flask will use to encrypt session data into the cookies.
• FETCH_PASSWORD: A random hex key to use that is needed to fetch all the survey data (such as user answers).
• MONGODB_URI, DATABASE_PATH and DATASETS_PATH: Please check Storage and Database

Storage and Database

Deployed on Surf SRAM

When Pictopercept is deployed in SURF SRAM, it uses two attached storages named db and pictopercept_datasets, which are attached in the server's /data/db/ and /data/pictopercept_datasets/, respectively. It will also use the mongo service that is defined in the docker-compose.yml file.

Because of this, the ENV used by the deployment (appart from the secret keys) will just be:

• MONGODB_URI: mongo:27017
• DATABASE_PATH: /data/db
• DATASETS_PATH: /data/pictopercept_datasets

Developing locally

Database

When developing Pictopercept locally, we need some extra external services such as MongoDB running and a dataset folder. There are two recommended options:

In memory via mongomock (non-persistent)

You can also pass the flag -debug when opening the server (via flask run --debug) to use the in-memory database. However, this is not persistent, and MongoDB mock service will restart on each flask run/update.

MongoDB using Docker service (persistent locally)

One easy way would just be using the Docker Compose service's MongoDB locally. To do this, you can run it and just set the MONGODB_URI to mongodb://localhost:27017/.

Database and Dataset folders

You will need to create two folders wherever you want and point both envs to those folders (DATABASE_PATH and DATASETS_PATH).