Computer Vision


Under Construction

I taught computer vision to high school students during 2 years. Instead of formal lectures approach we took on doing projects which included some computer vision techniques but weren't only about them. Below are the remnants of some of these projects:

  • NAO Reader App. This is was a project with Aldebaran Robotic's NAO robot series which have 2 frontal cameras and support Python. Using OpenCV and Tesseract libraries we were 'teaching' the robot to recognise text. In the end my students and I built an interactive "English language Robot Tutor" designed to teach children English spelling and pronounciation: a robot pronounces a random word from a list of known words, a child locates this word among cards, and shows to the robot which either recognizes the word as a correct one or not. Videos of the result be seen on youtube: Final result at some local School Fair, Recognition Part, Interaction Part. Some outdated chips of code can be found on my Github. I loved working with NAO series and would like to do this again one day. Or with any other robot
  • Brownian motion. This project was conducted a few years ago during a Summer School for high school students. We made a substance with glue and water (recipe kindly given to us by my Dept) to see and record via microscope Brownian motion. After the recording the video was processed with OpenCV and all round objects with their corresponding timing saved as .json to be visualized on the web with D3.js. Resulting visualization and its Code

I have a special interest in creating web animations and want to build some web world, like a game or some interacting environment. Below are a few of my projects.

  • Cherry Tree Quote. I have webcrawled BrainyQuote and similar sources for random quotes to include in a canvas based interactive environment. A user visits a page and cleans a misted glass to reveal a picture and a quote underneath. Github code.
  • DarQin: breathing and blinking Harlequin. This is a small animation of a little harlequin using JavaScript. Demo.

Currently there are a couple of projects I'm working on (aside the University's curriculum):

  • LanaLys A Twitter data based project that is aimed at finding patterns in a person's twitter data. As of now I work on language processing part and visualization of results. My test object of study is Trump. I'm going to finish it before May, 2018.
  • Christmas Crystal Ball. A Threejs webgl sphere which can be shaken with snow inside that moves according to the amplitude of shaking and a random congratulation and wishes appearing inside the sphere when snow falls back down. Github code as of now. I hope to finish it before Christmas.
  • 3RR (Postponed). A canvas game world of robots who live by the 3 Rules of Robotics as introduced in I. Asimov books
  • CVPlayRoom (Postponed). A small canvas game with elements of CV. User has to play in a canvas game using a real life object recognised from her or his web camera.
  • desTROY (Postponed) A 3D webGL world built employing some CV techniques. The concept: a person logs into the system and uploads a picture of someone despicable and hated. On the server the face is recognised and main features are extracted to create a villain resembling the uploaded picture. After that the user starts playing : she or he is constanly attacked by replicas of the villain and needs to destroy as many as possible to get the highest score.