BullsEye

About

BullsEye is a computer vision application for optical fiducial tracking on tangible tabletops. BullsEye is comparable in functionality to reacTIVision but offers significantly more accurate tracking of fiducials. BullsEye can serve events to a table application both through the TUIO protocol and through a Websocket connection.

The inner workings of BullsEye is detailed in the paper BullsEye: High-Precision Fiducial Tracking for Table-based Tangible Interaction.

BullsEye is freely available for academic and non-commercial use. If you intend to use it for a commercial product, please contact us.

System requirements

Microsoft Windows
Java version 1.8
A graphics card supporting OpenGL 3.2 or newer
A camera supported by DirectShow, uEye or MJPEG
An interactive table similar to what you would use for reacTIVision

Downloads

Setup

BullsEye requires a calibration of light, of camera distortion compensation and of the fiducial and finger tracking which is detailed below. BullsEye also contains embedded documentation available as tooltips on questionmarks in the UI.

Starting BullsEye

Download and extract BullsEye
Double-click BullsEye.jar, this should open a window looking like this

Calibration of light

Press 'Configure Cameras' and activate the camera(s) used for tracking (screenshot).
Now press 'Adjust Colour' and press 'reset' (screenshot).
Press 'Adjust limits'. This will activate the calibration of light. To calibrate BullsEye move a white object across the whole surface of the table effectively painting the image in BullsEye green. (screenshot, photo of white object on table).
Repeat for all cameras.
This should result in a nice image (screenshot)

Compensation for lens warp

If you do not need super accurate tracking simply press 'automatic' and skip this step. Otherwise:

First make sure either to have an external monitor connected to the computer driving the table, or connect to the computing remotely, as this step will involve covering the whole surface of the table up
Get a print of the calibration dot grid on a piece of paper with the same dimensions as the table surface. (photo)
Place it on the table and make sure it is firmly pressed down on the surface. (photo)
Press 'lens warp' in BullsEye, and make sure that all dots are visible. If not the paper is probably not pressed firmly enough towards the surface of the table. (screenshot)
Press 'calibrate lens warp'
You are now asked to click on an area in the image, try to press in a lighter area of the image for the best result.
BullsEye will now try to calibrate. (screenshot)
The calibration will probably not be perfect, so use 'expand' to let BullsEye automatically try to remove holes.
The resulting image should look as regular as possible (screenshot), if not try again by clicking at another starting point, or by fastening the paper even better.
Repeat for all cameras.

Placement

If only a single camera is used simply use autodetect. For setup of multiple cameras the two camera inputs have to be aligned.

Place some objects on the table that can be used as reference. E.g. some at the bottom and top of the table, and a long object crossing intersections (photo, screenshot)
Use arrow keys to move the grids, and pageup/pagedown to rotate.
Remember to update the size of the total grid. E.g. with two cameras stacked vertically, the height should be doubled.

Configure tracking

Press 'Configure tracking'
Select a quality, and note how it affects the framerate on your system. (screenshot)
Go to 'Fingers' and disable finger tracking
To configure tracking of tags, print a page from the BullsEye tag set and place it on the table, firmly pressed down on the surface. Make sure it is the only object on the table.
Goto 'Tags' and you should see the following: (screenshot)
Press 'autodetect' which should result in the following: (screenshot)
If you need finger tracking (experimental), go to 'Fingers' and adjust the parameters until fingers are tracked satisfactory.

Troubleshooting / FAQ

We use BullsEye in a lot of installations and it works for us on almost all of the hardware we have tried. That said, we do not have access to all of the hardware in the world to test it with. Sometimes things just do not work as expected and BullsEye will perform badly or not work at all.
This short list is a compilation of the most common problems:

My DirectShow camera is not in the list, it is not detected

If the camera works in other programs on the same machine make sure that the camera driver and BullsEye are running either 32-bit or 64-bit together. Most Java installations are 64bit while some camera drivers (like the ps3eye) are only 32-bit. Download a 32-bit version of Java in that case.

Camera is all white or black regardless of what I do, help?

If this happens even with "Passthrough"-mode in the light-calibration step then this type of error is usually GPU-related, maybe your GPU or graphics driver does not support the needed versions of OpenGL and the GLSL shaders used by BullsEye.
If you are on a laptop or multi-GPU machine make sure that it is set up to allow Java to use the powerful discrete graphics card and not a low-power built-in one.
Make sure you are running the latest drivers from the GPU manufacturer's website and that the card is modern enough.