Does Telraam respect privacy and confidentiality?
The Telraam camera is focused on the street to count pedestrians, cyclists, cars and heavy traffic.
Telraam processes the camera images immediately. The images from the camera are never stored. The camera also films in low resolution so it does not recognize faces or license plates.
The device is designed so that there is no way for the camera images themselves to be viewed (not by the owner of the Telraam device, but also not by third parties). The camera images are visible only when the Telraam is installed, only to the user (in order to properly aim the camera), and for a period of up to 10 minutes.
The data collected by the Telraam is transmitted wirelessly to a central database where it is further processed. The results of all the Telraam devices are freely available to everyone at www.telraam.net.
In a prior opinion sought, the Data Protection Authority (GBA-APD in Belgium) had no objections to this method of processing camera images (instantaneous and local, as opposed to retaining images & forwarding them to a central database for central processing as with traditional camera systems).
From the people who register to Telraam, we ask them to fill in some personal data (name, address and e-mail address). Before proceeding with this, they sign an informed consent, on what happens with these data, among others. This personal data is in no way shared with third parties. To share the Telraam data, we use road segment, and not specific address data.
All intellectual rights relating to the traffic counts collected and any database into which they are incorporated by Telraam will belong to the Telraam consortium. To the extent necessary, the participant's consent shall constitute an unconditional, irrevocable and royalty-free assignment of any other right or claim the participant may have thereto by reason of his/her participation. For his/her participation, the participant acquires a non-exclusive personal use right to all data produced by his/her own Telraam device and can consult and download them via the personalized dashboard.
The camera legislation is not applicable to Telraam (art 3 chapter II of the law of 21 March 2007 on surveillance cameras):
"This law applies to the installation and use of surveillance cameras in the places referred to in article 2, with the aim of:
1° to prevent, determine or detect crimes against persons or property;
2° to prevent, determine or detect nuisance within the meaning of article 135 of the new municipal law, to check compliance with municipal regulations or to maintain public order."
This is not the purpose of Telraam so the law does not apply to Telraam.
Data Protection Authority's (GBA-APD) response:
"Based on the additional information provided in your email, the use of the cameras as you have explained it does not appear to us to involve any processing of personal data."
The additional information referred to are:
- The camera is installed on private property (i.e. in a house) and is aimed at the street.
- The camera is fixed aimed at the street.
- The camera is not actually filming, in the sense that the images are processed immediately. There is no way to access the images themselves. Technically, processing is done as follows:
- The camera sends image (effective image of the street) instantaneously to the processing unit (a raspberry pi, a mini-computer), which is physically connected to the camera with a cable of 5-10cm.
- In the mini-computer, the images are processed immediately for object detection and stores only the following information about these objects: size, speed and location of the object on the image (top/bottom).
- The images of the camera itself are actually not captured but immediately converted and are also not visible anywhere.
The information about object properties is forwarded to a central database for a 2nd processing where object properties are translated into vehicles (e.g. object with size 5688 pixels with pixel speed 24 pixels/second is a car, object with size 700 pixels with pixel speed 10 pixels/second is a bicycle, etc...).
If you consider the whole camera & local processing, the input comes from the camera (basically frequent images in color spectrum per pixel), interpretation of the input happens by detecting general properties of objects (pixel size,...). In this sense, no images are actually registered, only the properties of the images relevant to us (object detection and some properties) are filtered out. So the camera counts the number of objects passing in front of it.
There are 2 essential elements here:
The images themselves are processed locally and immediately. In other words, it is impossible to consult the images of the camera directly (the device would simply not work if that were the case by the way, weighing too little computing power...)
The processed data are generic and are not personal data: no license plates, no faces, no characteristics of persons,...