This Program can be used to carry out paper based surveys.

The questionnaire is designed using either OpenOffice/LibreOffice and then exported to PDF. SDAPS then uses both the ODT document and the PDF file to figure out what questions are asked, and where checkboxes and freeform fields are placed. Another great way to use SDAPS is together with its LaTeX class. It allows you to create questionnaires very easily and is tightly integrated into the SDAPS main program.

After this, the program can create an arbitrary number of (unique) questionnaires that can be printed and handed out. After being filled out, you just scan them in, let sdaps run over them, and let it create a report with the results.

Depending on what part of SDAPS you use, different dependencies are required.

general (including recognize):

• Python 2.7
• distutils and distutils-extra
• python-cairo (including development files)
• libtiff (including development files)
• pkg-config
• python-zbar for "code128" and "qr" styles
• python development files

graphical user interface (gui):

• GTK+ and introspection data
• python-gi

reportlab based reports (report):

• reportlab
• Python Imaging Library (PIL)

ODT based questionnaires (setup/stamp):

• reportlab (for barcode rendering)
• pdftk or pyPdf (pdftk is much faster if you need questionnaire ids)
• python-pdftools

LaTeX based questionnaires (setup_tex/stamp):

• pdflatex and packages:
  • PGF/TikZ
  • translator (part of beamer)
  • and more

LaTeX based reports:

• pdflatex and packages:
  • PGF/TikZ
  • translator (part of beamer)
• siunitx

Import of other image formats (convert, add --convert):

• python-opencv
• Poppler and introspection data
• python-gi

Debug output (annotate):

• Poppler and introspection data
• python-gi

You can install sdaps using "./setup.py install". The C extension will be compiled automatically, but of course you have to have all the dependencies installed for this to work.

As an alternative to installing sdaps it is also supported to run it without installation. To do this run "./setup.py build" to build the binary modules and translation. After this execute sdaps using the provided "sdaps.py" script in the toplevel directory.

Please run sdaps with "--help" after installing it for a list of commands. Also check the website http://sdaps.org for some examples.

The quality of the recognition in SDAPS is quite good in my experience. There is a certain amount of wrong detections, that mostly arise from people not checking or filling out the boxes correctly. For example:

• The cross is not inside the checkbox, but next to it
• People cross the same box multiple times
• People use very thick pens
• Filling out is not done properly

As you can see, most of the errors arise from the possibility to correct wrong marks by filling out checkboxes. SDAPS tries to be smart about this by using different heuristics to detect the case, but it is not foolproof.

Suggestions on how to decrease the error rate are of course welcome.

It can happen that SDAPS is not able to calculate the transformation matrix which transforms the pixel space of the image into the mm coordinate system used internally. If this happens the affected pages cannot be further analysed. It is usually possible to manually correct them using the GUI, but that can be quite tedious.

See also TROUBLESHOOTING for some more information.