All Categories Data interpretation Telraam indicates that some of the data have a "Poor Quality"? How to interpret it?

Telraam indicates that some of the data have a "Poor Quality"? How to interpret it?

Do any data appear as "poor quality" on the site? What does this mean? 

As illustrated in the graphic below, Telraam will occasionally indicate that some data are of poor quality. This article will explain what this means.

Poor-quality-EN.png

Several explanations are possible, but most of these are only expected with Telraam V1 units:

  • While without ground truth counts we cannot control the quality of the data on a segment-by-segment and hour-by-hour basis, we can make sure that if a Telraam device is having issues with counting (for either technical or network connection reasons), then we flag the corresponding problematic intervals. In a one hour time interval, the recorded data will appear as "Poor Quality" if the corresponding Telraam unit has been active for less than 30 minutes (50% of the interval). So in case of a temporary power or network outage, or inadequate light levels (for example if the Sun rises at 6:40 am, or sets at 7:20 pm, leaving only 20 minutes of useful data for a one hour period), Telraam will show that the data for that period might be biassed and therefore of "Poor Quality". It is important to note that data with the “Poor Quality” label might be still quite accurate, assuming the traffic flow during the period was relatively constant and the volume was not too low. The reason for this is discussed in the next point.

  • The data displayed on the Telraam website (and accessible via the API) is always corrected for any kind of device down-time, in order to better approximate the actual traffic levels by avoiding the introduction of a time-dependent bias. This is especially important for data coming from Telraam V1 devices, which - by design - are only monitoring the traffic in 70-75% of the time (while the rest is spent on periodically capturing a background image and tracking the detected objects). When the actively observed period is less than 50% of the total time interval, then these corrected data will appear as "Poor quality". For example, if between 10 and 11 AM a Telraam has been active for only 20 minutes, the system will multiply the collected data by 3 to give the best approximation of the traffic during that hour.

  • Telraam S2 devices do not suffer from scheduled periodical down time, so as long as they are powered and there is a network connection, the only time we expect data to be “Poor Quality” is when the initial installation happens after 50% of an hour has already passed (so, e.g., at 35 past the hour). Then the first hour in our database will contain less than 30 minutes of counts, and therefore will be labelled as “Poor Quality”, but all following hours should be complete.

  • Moving objects not related to traffic flow perceived by Telraam V1 units can also affect the quality of the data. If for example a tree or an advertising flag is in constant movement in front of a Telraam V1, that creates a lot of spurious objects for the tracker algorithm that have to be filtered out, which takes an unusually long time. This load can extend the downtime from the typical 25-30% beyond 50%, which will result in the corresponding time intervals being labelled as “Poor Quality”. Such cases usually create consecutive intervals with “Poor Quality” data, since the root cause is continuously present. Times with strong wind might be the worst, as strong wind might cause trees or flags to move that otherwise appear static. Note that these spurious objects are filtered out correctly and are not counted.

  • In case of the Telraam V1 units, long periods of standing still traffic (traffic jams) might cause problems too. In this case background calculations might fail, as the stopped cars are perceived as static background, which will cause anomalous object detections as soon as the traffic starts to move, which in turn leads to a similar system load as the moving objects did, but in this case even the real objects will be counted incorrectly, leading to not only a “Poor Quality” flag, but actual bad quality data.

What should you do in case of 'poor data quality'?

1. Check on your camera image if there are any moving objects (e.g. trees) and remove them (if possible). Maybe a tree has grown in the spring and is covering (partly) the view? Follow these steps to check your image: Verify or update the camera view (1) or set a new ROI (S2).

2.  Add a sun reflective screen. You should have received this together with your Telraam device. It's a black paper you can cut and fold to cover your camera as a sun reflective. You can find it as well in the attachment below.

3. Poor data quality can be caused by poor internet connection. Follow these steps: My Telraam sometimes loses wifi connection .

4. Finally, if you have had your Telraam for some time, you should know that we are constantly trying to improve the Telraam system. The latest version (v11) was released in August 2021 and you can update your Telraam softare by clicking here: "Updating the Telraam software by yourself "

Was this article helpful?

Thanks for your feedback!