|
Anette von Kapri
RWTH Aachen meets Columbia University
|
|
|
While working on my master's project at
Columbia University,
New York, in Prof. Feiner's
Computer Graphics and User Interfaces Lab.
I am finishing my master's degree in Computer Science at the Rheinisch Westfaelische Technische
Hochschule
(RWTH)
Aachen in Germany. I will be graduating in March 2008.
|
|
Interests:
|
I am very interested in Virtual Reality and Computer Graphics. For over a year I worked at the
Virtual Reality Lab at the RWTH-Aachen University. Computer Graphics is my area of specialization
for the sudies, but for my master's thesis I worked in Augmented Reality, 3D User Interfaces and
also Computer Vision.
Besides the "graphical" part I am interested in Artficial Intelligence, for which I took different
classes during my studies abroad in Paris at the Ecole Nationale Superieure des Télécommunications
(ENST)
and a tutorial at my german university.
I also loved Prof. Hromkovic's courses about algorithmics for hard problems and complexity. He offered
me to do a PhD in
his department
at the ETH Zurich University (Swiss Federal Institute of Technology
Zurich).
|
|
Hobbies:
|
In my spare time I like to play the transverse flute, my intention is to give a big concert to the
Columbia Computer Scientists at the end of my stay with a good friend of mine who plays the piano,
we still have to exercise a lot...
Together with my father and my brothers, we travelled through Europe and also Egypt, which was
probably the most interesting and adventurous journey up to now. Besides, I like to go swimming and
biking and recently I discovered a new sport: Yoga.
Jogging is not my big passion but I participated at a "little" marathon (5,555 Meters) at an event
called
"Lousberglauf"
in Aachen 2004 and 2006
(running Anette),
where the whole city meets
to run or to watch and celebrate afterwards. Unfortunately for 5,555 Meters I needed 38 minutes
(over the two years though I improved of 2 minutes...).
I like to meet new people from different cultures and from all around the world, that is also a reason why I came to New York. The City is full of interesting people with completely different backgrounds
and ideas, everybody is open minded and wants to discover new things.
|
|
Projects:
|
These are some projects I worked on or participated in, during my studies, my work at the Virtual
Reality Lab and the master's thesis.
|
|
3D User Interfaces Lab, Columbia University:
|
For my master's thesis I have been and still am working in Augmented Reality and 3D User Interfaces.
For that I am implementing an annotation and information input system for an electronic field guide
for botanists to use in a field. To imagine the situation of use, think of a botanist who is walking
through the fields and he sees a plant unknown to him. He will take out his TabletPC, connect the
camera and film the plant. Life while inspecting the plant he can add information and annotations.
The added information will stick to the leaf even if it is moved around.
We distinguish between general information to a plant and specific areas on the leaf
which seem to be interesting. I want to give the user the possibility
to annotate a certain area on a leaf and save text and audio information for later inspections.
In this context I examine different interaction techniques with the leaf and the annotations,
to make the system easy to understand and fast to use. It will be interesting to see if the user is more
likely to add annotations to the plant on a still image which was frozen and where he can zoom in and out or
on the moving camera stream.
The botanists want to collect information for each leaf
and save it uniformly that each "information sheet" will look the same and the input process will
not diverge. So I am working on finding a menu structure which is in one way intuitive for
the botanist, easy to understand and on the other hand powerful for diverse plant data.
In this context I also explored, if it is possible to track a leaf and calculate its 3D position
and orientation. In Augmented Reality a system called ARToolKit is used to put virtual objects in relation
to real objects, by placing predefined trackable squares in the real world and then positioning
the virtual objects in relation to the calculated 3D position. I tried to find a way to track a leaf by itself through a vision based feature tracker called KLT Tracker. The idea was to detect feature points
on the surface of the leaf and through their relative positions to each other calculate the orientation and position of the
leaf, but this feature tracker was not stable enough for my purposes.
|
|
Computer Graphics Lab, RWTH Aachen:
|
In a student project
I worked on
building a low cost vision based 3D light scanner at Prof. Kobbelt's Computer
Graphics Lab.
There exist laser scanners which project a laser ray onto a real object and knowing the position and
orientation of the camera which captures an image, the 3D point of the laser ray on the object can be calculated
through triangulation.
The paper
"A low cost 3D scanner based on structured light"
by C. Rocchini, P. Cignoni, C. Montani, P. Pingi and R. Scopign
proposes a cheaper and healthier method to scan a real object, by projecting a red, green and
blue pattern on it and then calculating the related points.
My group was in charge of the Detector Calibration, Image Processing and
Triangulation. This means basically that our task was to create a pattern of alternate red
and blue and inbetween a green line, which should be projected on the object to be
scanned. The green line lying on the surface of the object should be extracted and
knowing the 3D plane of the green light plane and the camera position, the 3D points
of that green line on the object could be calculated through trianglation.
The difficult part in this project was to find the correspondences between a green light plane and its
projection on the object.
|
|
Virtual Reality Lab, RWTH Aachen:
|
As a student researcher I worked in Virtual Reality for more than one year. I loved the work there.
The code was very organized and written so that it could be reused easily.
My advisor
was great, he taught me a lot.
I was programming in C++ and VRML also modelling virtual objects in 3D Studio Max and AC3D.
During my time there I also worked on 3D sounds and adding it into the virtual environment.
But mainly I concentrated on building virtual avatars on the example of fishes, I simulated their
behavior swimming and avoiding each other in an aquarium. Through collision
detection they could avoid each other and decide based on a probabilistic method in which direction
to swim next.
|
ENST, Paris:
Virtual Reality:
|
For seven months I stayed in Paris studying at Télécom Paris. I have a deep emotional relation to
France and Paris, when I was little my grandmother lived there and I stayed with her for a year. My
parents educated me biligually, french and german, I have a lot of connections to France, a lot of
friends there. My study abroad was half motivated by visiting another very good university and half
through my "french" background.
In the Virtual Reality class we students worked on a project where we had to recreate a
fable of:
Jean de la Fontaine in a modern way in VRML. I chose the fable of
The Ant and the Cricket, where an ant is working all the summer
to build up provisions fo rte winter and the cricket is just singing and enjoying his time. When the
winter comes the cricket does not have anything to eat and asks the ant to help him. The ant is
furious and says: "I have been working all the summer to have enough to eat during the winter while
you were dancing and singing, so now just continue!"
|
|
Collective Intelligence:
|
For the Collective Intelligence class my project partner and I, we worked on
implementing the mechanism of learning concepts and the association of these concepts to words
according to the thesis of Andrew Smith
"Evolving Communication through the inference of meaning".
Basically he was interested in the fact how a new language can be learned without having any
knowledge about it, like for little babies, who just learn the language over time through seeing in
which context the different words are used. So in our program two agents are in a world and they
don't speak the same language. Eventually one agents says a word related to an object which is
currently seen. The other agent saves a table of probablities for each object and each word,
depending on how often he heard this word while the specific object was in his sight. After a while
he can be almost certain what the different names of the objects are.
|
|
Pattern Recognition:
|
For Pattern Recognition we developed a Hand-Writing Identification algorithm based on sensor vectors.
Our task was to identify handwritten letters.
Instead of considering all pixels of an image of a handwritten letter, another possibility is to
shoot vectors from the margin of the image to the opposite site and see when this vector reaches the
first pixel of the letter. Based on the information you get from the distances of these vectors to
the letter, it can
be recognized. We experimented how the amount of vectors and their starting positions would
influence the recognition of the letters.
|
|
|
