ANR TypoRef

This project was a school project of 3 months to shape our software engineering and Java lessons. We had the choice between 3 subjects, and we chose the hardest but also the most interesting. The teacher that gave this subject needed an app for the researchers of the CESR of Tours. For their researches they have a database of drop caps dated from Renaissance. When they find new drop caps, they need to compare it to the ones they already have, in order to date it, find its designer... Our teacher had to develop an AI of shape recognition to make the comparison process easier. Hence he asked us to make the software that will host his AI and will be used by the CESR researchers to classify their drop caps images. The fact that this project is an ANR project makes it impressive because it's not just an inner project of the University of Tours, it's a national project, so it was a chance to be able to work on such a project.

We sliced the project in 3 parts :

• First part was all the app design, so we made use case diagrams, sequence diagrams and in the end a package and a class diagram. We also thought about the database for our app.
• The second part was supposed to be about making a clean app by doing a Javadoc and writing the unit tests of our methods. But, remember when I said the researchers already had a database of drop caps ? Well they gave us the database and it was an absolute mess. So we had to do a huge work of normalization in order to extract the interesting data. We even made Java scripts only in order to fetch some troublesome data that were concatenate with other ones. Because of that we rushed the unit test and had to do a lot of debugging in third part.
• For this third and last part we had to actually write all the application. We wrote a lot of code and rushed the design because we struggled a bit with the JavaFX interface. In the end the app was fully functional, but well it was... ugly.

So the initial database looked like this :

And we came up with this database which is definitely less confusing :

Here are some key visuals of the app :

For this last picture, as the drop caps can be described by chosen key words by the researchers, we thought that it might be a good idea to do a word cloud of the keywords in order to visualize which keyword is the most used. We did that with a small Python script.