The Thesis

Your thesis, be it Bachelor’s or Master’s, is the most important work you will do during your studies. It proves to yourself, fellow students, friends, the university and everyone else that you deserve the title ‘computer scientist’ or ‘software engineer’. It is the final culmination of the two or three years your studies. Use every skill, tool, experience and piece of knowledge you have gained so far to make a great thesis!

You can only do it once in your life!

Most importantly, it is your thesis. You will be the one responsible for it. You are the one to reap the benefits and the one to proudly call yourself the author of it. This is why the topic of the thesis should be something you feel that is right for you. Something that you really want to be good at and work with after your studies as well. Perhaps not with exactly the same thing, but definitely with something similar. Because, you see, after writing a thesis, you become very skilled and knowledgeable in the particular topic. It becomes the road you have traveled on and know well.

The choice of a supervisor can depend on many aspects. The supervisor is not your client or boss though. Neither is your supervisor a lecturer or a practice session instructor who would give you a grade afterwards. Do not expect the supervisor to tell you exactly what to do and when to do it. Those questions are for you to solve. Your work is for you to keep track of. Your supervisor is more like a co-worker at a coffee table. Together we can come up with some great ideas or try to solve difficult problems. We also care about how you are doing and will give you a cheer if you are feeling down. Naturally we want to help you and give you feedback on your results, but all the choices are and will always be yours to make. It is your thesis and you will be the one defending it afterwards. You need to be confident of your choices in your thesis. All of them.

Be sure to pick a supervisor that you like to get along with. Someone who also has some experience in the field that interests you. You should be able to discuss ideas and understand each other. Someone whose support you find the most useful.

The best topics are often the ones that you come up with yourself. Remember that doing the degree was your choice. What was it you wanted to learn from your studies? Take that idea and make it your topic! No one else should tell you what job to get in the future, which house to buy or how many pets you need to have. Coming up with and choosing a topic is just like those choices. It will affect you and your future – what is the area you want to excel in?
Among other values, having a degree means that you have shown to have independent thought in identifying problems and setting goals plus the skills, creativity and willpower to provide solutions, follow through and make the world better!

Be sure to check out the computer graphics theses from previous years.

Frequently Asked Questions

1) What is a thesis?
Please read the thesis regulations PDF (in Estonian, in English).

2) Can I make a computer game as a thesis?
Yes!
It does need to be unique though, not just a clone of an existing game. It also needs to satisfy the rules for the “software solution” thesis type. Among other things you need to: describe similar games, say why your game is different, explain the design choices, describe the more complicated algorithms, test your game on the players, improve it if necessary and show that it is a great game!
If you make a game, which will have an Overwhelmingly Positive score on Steam, thousands of players and manage to write up in your thesis what you did right, then it will be awesome!

3) What should I reference in my thesis?
It depends on the specific topic. Generally, you should find good books and science articles that support your topic. If you are a slow reader, then the earlier you start, the better. ACM SIGGRAPH is a nice source for computer graphics articles (library link).
It is fine to reference non-academic sources too, but it should not be only those. Otherwise, if a source is non-academic, but relevant for your work, then do reference it, of course.

4) It is now February… I need to do a thesis… can you help?
Where were you in December? Do you not enjoy your sleep?

5) How do I know if my topic is suitable?
Do you feel confident to defend it according to the thesis regulations? If you have not read the regulations yet, then please do. Then think about, which type your thesis is of. Then think, how does your topic satisfy the requirements for that specific type.

In almost all the cases, you should be able to answer right off the bat:

  • To whom is your thesis meant?
  • What is the unique value in your thesis?
  • What are the similar / alternative solutions in the world?
  • Do you utilize enough of computer science?
  • Is the foreseen work worth 9 (BSc) / 30 (MSc) credits?

If you can adequately explain that, then your topic is probably suitable. Come talk to us! 🙂

 

Thesis Topics

If you really can not find a topic on your own, then here is a list of potential topics we offer.

 

Delta Building Visualizations

Supervisor(s): Raimond Tunnel
Degree: BSc or MSc

The institute wants you to create some nice visualizations to be displayed on big screens around the new Delta building. The topic is quite loose and probably would be connected with people developing the hardware solutions as well. One idea is that we could put sensors around the Delta building and then visualize how many people are in different areas (with a 3D stylized model of the building and silhouettes of people / students).
But if you have an idea what to put on some huge screens, which would illustrate computer science and computer graphcis, then all ideas are welcomed. This topic can also be done in a small team with interested students and can have some 1-3 related theses. So if you want to do something with a friend, this is a perfect opportunity.

valve-the-lab-virtual-reality

Wolf3D Topics

Supervisor(s): Margus Luik
Degree: BSc or MSc

Wolf3D is a company in Tallinn. They do quick scans of people with a mobile app and then use the results as avatars in 3D games. There are a number of things they want to improve in their app and are looking for smart students who could get experience with several bleeding edge algorithms. So those topics would be about taking an algorithm, seeing if there are alternatives, implementing the algorithm and validating that it works or does not work in Wolf3D-s software. Each algorithm is sufficient for one thesis. BSc and MSc distinction comes from the amount and depth of the work.
These topics can also involve paid work, which can further contribute towards the Practical Training in Informatics, if that is in your specific curriculum.

In the order of importance:

  1. Coupled 3D Reconstruction of Sparse Facial Hair and Skin
    https://www.disneyresearch.com/project/coupled-3d-reconstruction-of-sparse-facial-hair-and-skin/
    How to detect and recreate a beard, eyelashes and eyebrows?
  2. Image Style Transfer Using Convolutional Neural Networks
    http://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Gatys_Image_Style_Transfer_CVPR_2016_paper.pdf
    How to transfer a certain aesthetic style automatically on a texture or a 3D mesh. For example making the human’s avatar resemble a character from the Simpsons.
  3. Virtual Faces
    http://graphics.uni-bielefeld.de/research/faces/ from Figure 5 + refs
    Making a realistic avatar into a stylized one.
  4. Robust face landmark estimation under occlusion
    http://www.vision.caltech.edu/xpburgos/ICCV13/
    Detecting hidden features in photos and videos.
  5. Photorealistic Facial Texture Inference Using Deep Neural Networks
    https://arxiv.org/pdf/1612.00523v1.pdf
    Generating a texture from one image. This is probably a huge topic / work.
  6. A Morphable Model For The Synthesis Of 3D Faces
    https://www.cs.cmu.edu/~efros/courses/AP06/Papers/Blanz-siggraph-99.pdf
    Creating a 3D model from a database and an image.

 

Intelligent Language Teacher

Supervisor(s): Jaanus Jaggo
Degree: BSc or MSc
Tools: Javascript, HTML5, Amazone Services

I’m currently developing a web-based natural language learning application, that turns your favorite book into grammar learning exercises. The future goal is to add an intelligent teacher that studies the user progress and determines right exercises at the right time.

novelstyle

Hallucinations from Overweighting of Perceptual Priors

Supervisor(s): Madis Vasser, Jaan Aru
Contact: madis.vasser@ut.ee, jaan.aru@gmail.com
Degree: BSc / MSc
Technology: HTC Vive + Unity or Unreal

The thesis studies the top-down cognitive processes on perception and prediction by creating a uniformly gray world where the study subject must look around in order to spot tiny critters moving around. The aim is to induce the perception of stimuli even when none actually exist. It’s also important to track the users movement and responses in VR.

Probing the Internal World Model of the Brain

Supervisor(s): Madis Vasser, Jaan Aru
Contact: madis.vasser@ut.ee, jaan.aru@gmail.com
Degree: BSc / MSc
Technology: HTC Vive + Unity or Unreal

The thesis studies the internal world model of the brain by putting subjects in a VR environment consisting of visual white noise. The task of the subject is to report the exact moment when they start perceiving a generic environment from the noise. The environment fades in and fades out – the aim is to find the upper and lower bounds of such perception. We expect strong top down effects to occur, meaning that subject might perceive complex scenes even when none is presented.