I’ve posted on the Telraam Facebook page about this, but it’s worth repeating here.
There’s some configuration built into he v1 Telraam code that stops it counting when there’s not enough light, but also when there’s too much light. Typically most users won’t see this - devices are pointing at the street & buildings etc.
Out of a small deployment I’ve now seen two locations where we’ve hit this issue. It looks like having open skies in the background (without buildings or trees) is enough to cause the conditions where there is ‘too much light’ - even if the camera image is not really including much in the way of sky. If this issue occurs then device stops counting untill light levels reduce. This clearly is a pain if you want your Telraam to count consistently. Most users will see this as gaps in counting during the day - likely ‘poor quality’ where counting took place & did not take place in a given hour, and complete gaps were no counting happened. Clearly there’ll be a pattern of sunny days… Both instances where I’ve seen this are in northern England (UK) - near Manchester (yes… we do get sunshine here ). The problem only seems to happen after early March and before the end of October. Both instances are west facing (we don’t have many south or east facing devices), and seem to be affected from about midday. Other symptoms are poor quality count (cars getting counted as 2-wheels etc). To identify the issue I installed the dev software version & connected via SSH - I think there was a way to display the running code, and a ‘too much light - counting stopped’ type message pops up, though this was several years ago. Other ways are to monitor current draw of the device either using a socket meter or a USB current meter - current will drop from >2A to around 0.5A when counting stops, and to use the Telraam website along with any sunlight records. I raised a ticket but there’s been no software fix (the upper light limit must be there for a reason, but the code does not explain why). The fix I’ve used is to put a camera filter in front of the Pi Cam lens - I use polarising filters - these reduce the light level to around 50% - there’s a small drop in the daily count duration (at the start and end of the day, but it’s typically around 10 minutes each day. Previously I’ve taken the filter out in October and replaced in March, but I forgot to do this last year!.
There’s an example of the issue here - I think this is when I put a filter in place (e.g. you can see the difference in count quality) - this was during the period of fantastic (consistently sunny) weather we got in the UK at the start of Covid Lock down: period showing ‘too much light’ affecting counting, and installation of a filter (workaround / fix)
(I bought cheap used camera polarising filters off eBay for about 3-4 Euro, and then carefully ‘unscrewed’ the metal frame to leave just the glass disc). Any form of polarising filter should work, as should any filter - it’s simply reducing the light by around 1 to 1.5 f-stops. Polarising 3D cinema glasses lenses should work if you have some & they are flat. I stick to the window glass using 3M VHD tape which tends not to leave a residue.
Hope this helps someone (likely any issues will start again from late February / early March)
(having the paper cover on / off did not make any noticeable difference, I’ve only seen the issue when the camera is pointing between 0 - 90 degrees to the sun - e…g not when the sun is behind the camera)