Segmentation Painter Tool

Final project by

Rinat Gotsulsky & Roni Rabi

The ability of people to perceive structure in images exists apart from both the perception of 3D and from the recognition of familiar objects.
The action of organize the data even when we have no idea about the picture meaning, is important to the Computational Vision and can help to analyze models (known & unknown).
We focus on Segmentation- one of its issues. The approach is to divide a single picture into components, objects, text, shading, colors etc. We are trying to imitate the evolution solution to this approach, becouse Splitting an image into segments can be an extremely hard problem.
There are no unique definition and one picture can be explicit in billions ways.

Take a seat !

Do you think that what we see is just simply as that ?.
We made this project to deal with Segmentation method of Perceptual Organization.
Our goal to build "Segmentation Painter Tool".
The main idea is to test people about there way to divide a single picture to segmentation by color, shape, text etc... and get a clue what present in it and what object deferent from other.

Definitions

• Simple picture is present on "Segmentation Painter Tool".
• Pepole from defrent knowledge invited to try and tag pictures to models and objects.
• The decision where and what was in there hends.

Programing the code

We developed a toolkit enable people to separate picture to segments.
It enable us to learn about this process and later to apply this approach in computers systems.

The painter tool is on Java platform with the ability to load pictures from deferent types and drawing on them by fallow the mouse trucker.

The tests through use of program:

1. Each student draw on pictures alone without any clue aboute the target.
2. Only requirement data transfer to examinees.

A few examples for the outputs of Deferent knowledge students:
Students of :

.................Orginal pic..........................CS+Vision............................CS only..................Not from "TEVA" department

Segmentation :

Discuss the results vis-a-vis your goals and make conclusions.

1. In our small research there is no one explanation for a paint.
2. The segmentation was similar in all of the groups and maybe its from the fact that we see and investigate picture all the time and not because experience and learning of computational vision.
3. Its not obvious to segment picture – its not a well done algorithm.

Program SegmentPainter :

Discuss the results vis-a-vis your goals and make conclusions.

1. There are few must features that need to be added to our toolkit.
2. The mousses in computers lab must be replaced to better ones (optics).
3. Segmentation is very difficult to draw.
4. Need a good qualifications in painting to draw a segment picture.

• PDF presentation
• Oral presentation slides (PPT)
• Zipped files

1. Google.
2. Grader/TA- Andrey Rubshtein code.