This tool is available through the Rich Content Editor in Canvas, and allows users to extract text from images, either uploaded through the tool, or taken from the user's course files. It has two options for OCR models, Tesseract (Open Source) and Vision API (Google). You will need an API key to use the latter.
It was built from the LTI 1.3 Flask Template.
First you will need to clone the repo, and create the environment file from the template.
git clone https://github.com/dgwn/cyclops
cd cyclops
cp .env.template .env
Save your Google API JSON as a file api.json in your project directory and make sure it is .gitignore'd. Be mindful Docker currently copies this file into the container, so take caution not to share your image with anybody.
In this simple framework all the variables are preset, but for production you will want to edit the .env environment variables DEBUG and SECRET_KEY.
We use Docker-Compose to build and run our services.
docker compose build
docker compose up -dAfter Docker builds and starts the services, you will run the migration commands to create the database.
docker compose exec lti flask db upgradeThe database which will hold your LTI1.3 credentials has now been created. It's now time to generate the LTI 1.3 keys for LMS authentication:
docker compose run lti python generate_keys.pyThis script will output directions to follow to generate the Client ID and Deployment ID. You can find further documentation here: https://github.com/dmitry-viskov/pylti1.3/wiki/Configure-Canvas-as-LTI-1.3-Platform
The tool will now be running at: http://127.0.0.1:8000/cyclops/ and available via the course navigation from the account or course you installed the tool into.
Dmitry Viskov for the pylti1p3 python library.
Instructure for their LMS: Canvas
IMS Global for defining the LTI standards.
For more info on Google Cloud Vision API see here.
For documentation on Tesseract see here.