Calibration of your Telraam device takes up to 7 to 21 days
If you have the feeling the ratio of cars to large vehicles seems off in the first days and hours after installing your Telraam, this can be explained by the fact that the Telraam requires a calibration period of 7 to 21 days.
The classification of all objects is done based on two parameters, both being speed and distance independent: the fullness value and axis ratio. Based on these two parameters, a distinction is initially made between the category of pedestrians, two-wheelers and cars+large vehicles. The article "How are the different types of vehicles classified and differentiated? " talks about this in more detail. The distinction between cars and large vehicles is then done through a cut-off value. This value is an automatic value during the first week, and then a cut-off value is calculated specifically for your Census window in your house on your street during the calibration period. This specific cut-off point provides a better difference between these two types of vehicles that your Telraam counts on your street.
The calibration happens automatically and the data improves automatically the longer your Telraam is active. This takes at least one week. The shorter the days (and therefore less data), the longer this takes with a maximum of three weeks.
You can assume the sum of the number of cars and large vehicles usually to be correct, but in some cases the distinction between cars and large vehicles is still not correct after the calibration period. The number of cars is then underestimated and the number of large vehicles overestimated, or vice versa. One reason for this may be poor lighting conditions causing long shadows of cars that are then counted as large vehicles. The dividing point between a car and a large vehicle is now roughly around vans. However, it is possible for large cars (SUV, pickup,...) to be counted as large vehicles and small, low vans to be counted as cars.
We therefore ask you to always be careful when using large vehicle count data by Telraam. These data are not as accurate as Telraam data for the sum of cars and large vehicles.
After the initial calibration, further calibrations will take place automatically in regular intervals, typically two times a month. These are always based on the observed properties of the most recent two weeks' worth of data (or looking back at longer time intervals if necessary to have at least 10000 vehicles in the sample). This way Telraam can continuously fine-tune the cut-off value separating cars from heavy vehicles.
Due to technological limitations - in some specific circumstances, e.g., in streets aligned such that shadows cast by passing-by vehicles in sunny weather are often seen leading or trailing the vehicles themselves (artificially enlarging their perceived size) - it can happen that consecutive calibrations result in significantly different cut-off values. This mainly depends on the prolonged presence or absence of shadows, therefore the prevailing weather during and after the calibration period. The highest chance of this happening is when the weather suddenly changes following two weeks of constant overcast or constant sunny weather. This can lead to a sudden jump in the ratio of heavy vehicles and cars after the new calibration. If you see a sudden change in heavy vehicle numbers, without a change in the total number of cars and heavy vehicles, independent of the time of day, and which is maintained for several days, then there is a good chance that this is what is happening. Our Telraam S2 device is built using a different technology and it is largely unaffected by these moving shadows.
