In 2012-03 I had the chance to test Vicon Revue (v1) for Lifelogging. The Vicon Revue is the commercial product of the Microsoft SenseCam.

Vicon Revue is a neck worn digital camera with sensors. It takes (small) photographs whenever the sensors get a pattern that causes the firmware to make a snapshot. With the set of photographs you can see where you have been while wearing the device. Using those photographs for generating a small video helps people remembering their life. This has come to attention in the research of memory impairments.

I analyzed the Adobe AIR based software for downloading the data from the Vicon Revue and for reviewing the photographs later on. And I realized that I need an alternative software method to access and review the data: Adobe AIR will not be around forever and I want to avoid any lock-in effect.

During the analyzing phase, I found out that the original JPEG photographs do have correct Exif metadata included. Additionally, in every image folder, there is a file named «SENSOR.CSV» in CSV format which contains the sensor data and their time stamps.

After Vicon Revue Desktop moves the data from the device to your computer, the Exif metadata gets overwritten with the time stamps of the import process. Unfortunately, this is where you lose the time information within the photographs.

To get to the original times you have to either parse the «SENSOR.CSV» or the additionally created SQLite v3 database. The SQLite database contains all values from all «SENSOR.CSV» files and anything you add within Vicon Revue Desktop (notes, tags, titles, …).

But I want to use the photographs outside of Vicon Revue Desktop too!

Therefore I found the standard behavior a bit «rude». My solution I present here adds metadata like the time-stamps to the photo files itself and to the corresponding «SENSOR.CSV» files as well. When you do this before the import process of Vicon Revue Desktop, you keep the time information (not within Exif but) within the file names at least.

… as in v0.1 on 2012-03-23:

Usage:
         vksensecam-getphotos.py [options]

This script does has *two* different modes of operation depending
on the format of the given SENSOR.CSV:

MODE 1:
This script copies sensor data and photographs from a "Microsoft
SenseCam" or "Vicon Revue v1" device to your hard disk and:
   * creates a download folder on your hard disk
     like e.g. "/configuredpath/2012-03-23T21.18_SenseCam_data"
   * copies SENSOR.CSV and photographs to this download folder
   * add meta-data such as ISO time-stamps to the file names of photos
     like e.g. "2012-03-21T10.41.48_SenseCam_IR_00001810.JPG"
   * re-writes the SENSOR.CSV according to the new file names
   * stores original SENSOR.CSV as a backup
   * Note: EXIF meta-data of the photos is kept unchanged in the 
           photographs. Usually they are fine (reflect their exposure
           time-stamp and such).

No data is changed on the device itself! You can use other software
like Vicon Revue Desktop as well afterwards.

Example invocations on GNU/Linux or Mac OS X:
  vksensecam-getphotos.py --simulate    ... please use this if unsure!
  vksensecam-getphotos.py --verbose     ... to get all the noise details
                                            what is going on
  vksensecam-getphotos.py               ... this is all you need when 
                                            everything is set up

MODE 2:
If the given SENSOR.CSV is recognized having the format of an already
downloaded file (Vicon Revue Desktop changes date-stamp separator
characters to slashes), the tool assumes an already downloaded
data-set. In this case it does (in the same folder as the SENSOR.CSV):
   * rename photo file names like described above
   * changes entries in SENSOR.CSV according to the new file names
   * saves original SENSOR.CSV as a backup
   * Note: EXIF meta-data is already likely overwritten by Vicon 
           Revue Desktop before

Example invocations on GNU/Linux or Mac OS X:
  vksensecam-getphotos.py -f /path/to/sensorfile --simulate
             ... please use this if unsure!
  vksensecam-getphotos.py -f /path/to/sensorfile
             ... this is all you need when everything is set up

If unsure, please use the simulate option!
Basic configuration is done in the first section of the script.

  :copyright:  (c) 2012 by Karl Voit <[email protected]>
  :license:    GPL v2 or any later version
  :bugreports: <[email protected]>


Options:
  -h, --help            show this help message and exit
  -f PATH, --folder=PATH
                        (optional) override default folder for SENSOR.CSV
                        which is "/media/A295-005B/DATA"
  -v, --verbose         enable verbose mode
  -q, --quiet           do not output anything but just errors on console
  -s, --simulate        do not change any file on disk, just report what would
                        happen
  --version             display version and exit

The short shell script lists the files which are “interesting” to the user:

• images that were taken because of pressing the “take a photo”-button
• images that were renamed from their original file name

This gives a quick overview over a set of SenseCam images if the user has already screened the images and renamed important image files.

This sort shell script uses a command called “open” (exists on Mac OS X and can easily be done for GNU/Linux) to open all images that were invoked manually. If you close the image viewer, the next one opens.

This is very handy to quickly find and rename images which were shot on purpose - using the button.

• Wiki: http://www.clarity-centre.org/sensecamwiki/index.php/Main_Page
• Tech Specs: http://www.clarity-centre.org/sensecamwiki/index.php/Technical_specifications
• more technical details: http://www.clarity-centre.org/claritywiki/images/7/72/SenseCam_User_Guide_v1.4.pdf

This documentation is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.

The script file is licenced under GPL v3 or newer.