Social Expression Toolkit
(aka Attention Meter)

How can we monitor social expressions and provide adaptive feedback?

Jackie Lee, Heymian Wong, Anna Shcherbina
Affective Computing Group, MIT MEDIA LAB

Update (Aug. '11):
All source code is now available at GitHub.


Social Expression Toolkit allows researchers to implement and prototype ideas from social situations. We implement a real-time facial interaction platform (written in Python) based on Intel OpenCV face detect function.

Attention Meter v0.0.1 for Adobe Flash by Jackie Lee (Jan '10)

Attention Meter can be used remotely. This is the latest version (Jan. 2010) using Adobe Flash (please enable your camera to test it.)

Current Build:
We upgrade to OpenCV 2.1 with latest Mac OS 10.6.4. see below (I haven't tested it under Window 7.)

Older Builds:
Attention Meter is written in Python 2.5 with OpenCV and CVtypes. We also have an online version implemented in Adobe Flash Actionscript 3.0.

Attention Meter v0.0.2 for Mac 10.6.4 by Jackie Lee (Oct. '10)

1) Download OpenCV 2.1 private binary build from
This private build of OpenCV 2.1 will save you hours of time making OpenCV to work with Python.

2) Download
3) $ python
4) Done! (Let me know if you got any problem)

Attention Meter v0.0.1 for Mac by Jackie Lee and Heymian Wong (Spring '09)

1)Download and install Python 2.5 (on OSX.4.11, I haven't tried OSX.10.5 yet.)
2)Install Xcode and X11 SDK (It gives you most packages you may need.)
3) Download opencv-1.0.0.tar.gz from and unzip it.
4) cd into 'opencv' and do following commands (detail instructions from Mac OS X OpenCV Port):
a. $ mkdir build; cd build
b. $ ../configure CPPFLAGS="-I/opt/local/include" LDFLAGS="-L/opt/local/lib"
(You may add --without-python if you have an old python-wrapper.)
c. $ make
d. $ sudo make install
5) Download and unzip
6) do $ python (You should be able to get it running at this point. Please let me know if you can't.)

Attention Meter v0.0.1 for Win by Jackie Lee and Heymian Wong (Spring '09)

1) Install Python 2.5
2) Unzip
3) run in command line > python

Attention Meter v0.0.1 for OLPC('XO')
by Anna Shcherbina (Jan '09)

1. From the terminal activity type:
$ yum install opencv opencv-devel opencv-python
(This should install all the needed forms of opencv)

2. Type:
$ wget /usr/lib/

3. Put the program files in the "/home" directory. So if you just start python in that directory you should be able to run the files.

**The key to installing opencv is this download: /usr/lib/

It is a fix that allows opencv to work on the XO. There is more information about the fix on this site:

I would like to thank Dr. Ted Selker and Jon Wetzel (MIT M. Eng.) for their valuable contribution to "Attention Meter" project. Previous versions can be download through here:

Attention Meter (in VC++) [source] [exe]
Face Interface (in VC++) [source] [exe]

(Last updated on August 10, 2011 )






Jackie Chia-Hsun Lee
ffective Computing
20 Ames ST. E15-443D
Cambridge, MA 02139, USA
+1.617.452.5627 (office)