Motion detection almost useless

[SPFLMF_288692412260538]

Is there any development happening on this? I also find I'm getting many false alerts due to clouds, trees, shadows, wind moving bushes etc. I've tweaked all the settings as recommended, masking off as much as possible but can see no way of masking shadows and light level changes without effectively masking the whole image thereby rendering the detection useless.

[p.potarzycki_297194982211769]

I have the same problem. If the sun suddenly sets behind a cloud and vice versa, the camera changes the image parameters and it is considered motion. The detection sensitivity and detection areas do not matter as the whole image suddenly brightens or darkens. If the camera itself changes the image parameters, it should not treat it as a movement (changing the appearance of two adjacent frames). On a sunny day, I get hundreds of fake notifications. It's flustering.

[tehsolace_306849681780895]

The engineers need to research into machine learning image comparisons. There are many open source solutions to start from.

The basic idea is you use your previous set of images to compute a model, and then that model can be used very swiftly to calculate just how similar (or different) the current frame is from that baseline model. If its X% different, a notification should be triggered. This X% is what the user would consider "sensitivity".

Here is an article about how to use TensorFlow in python to do this, but ML is available in most popular languages, including C# (ML.NET), Java, etc.
https://towardsdatascience.com/image-similarity-detection-in-action-with-tensorflow-2-0-b8d9a78b2509

[Peste_218048536527094]

@tehsolace

They do have ai detection but they are only implementing it on their new line of cameras leaving all their other customers with a useless detection. Like others here, if I put to sensitive, I get warnings left and right, if I put less sensitive, I miss certain events. It's really useless.

[florian granel_313129877487844]

Hallo zusammen,

ich habe 4 RLC-410 im Betrieb und bin wirklich enttäuscht von der nächtlichen Bewegungserkennung. Diese Nacht 150 Meldungen ohne dass etwas, außer vielleicht Mücken, zu erkennen ist. Setzt man den Wert hoch, erkennt er Personen, aber löst auch mindestens alle 5 Minuten Alarm aus, setzt man den Wert runter, keine Fehlalarme mehr, aber Personen werden auch nicht mehr erkannt.

Ich habe auch noch eine Reolink Argus PT in Gebrauch, dort funktioniert es einwandfrei, keine Fehlalarme trotz voller Sensitivität und alle Personen werden erkannt.

Wo liegt das Problem bei der RLC-410? Wird die Personenerkennung der neuen RLC-510A dieses Problem lösen, weil dann schicke ich die ganzen 410er wieder zurück.

[Cynthia_124785627824270]

Es tut mir ziemlich leid für das Problem, das Ihnen verursacht wurde.
Könnten Sie bitte eine E-Mail direkt an support@reolink.com senden.
Wir werden unser Bestes geben, um Ihnen bei der Lösung des Problems zu helfen.

[florian granel_313129877487844]

Habe mir die Kameras noch mal angesehen, denke es liegt am fehlenden PIR-Sensor (Bewegungserkennung über bewegende Wärmequellen) die Argus-Cams haben einen und alle POE-Cams nicht.

[MikeG_312186640117904]

Same problem with the RLC-520 as well. A firmware update to improve the motion detection would be most appreciated...especially given the same algorithm seemingly has been used across multiple cameras given the age of this thread.

I expected the large range of sensitivity control would allow for reducing false positives, but like others have mentioned, no matter how much tweaking is done, you simply cannot get a good balance. The sun moving in and out of the clouds triggers alerts constantly, the camera switching from day to night mode triggers an alert.....these are scenarios that should NOT produce motion alerts.

Please invest some time in updating the firmware..

[Sstabb_616908805689437]

big issues still in 2022. seems by now there would be some AI learning available.
Its really worthless having motion detection when it goes off every 5 -10 minutes.
Even goes off as the sun comes up.

I have been tweaking the settings for weeks, this is going to requires some better technology

[chris_180]

Is there any update to this issue? Passing clouds drive my cameras nuts. Turning an outdoor light on or off also triggers them. Even fog mist makes them go off. The Arlo cameras this system replaced never once went off in these situations.

[t haugaard81_638012253938178]

have the same problem. Detecting anything But people, cars, anomals and so on. Only get detektion from spiders, insekts, cloud, sun set og car frontlights at night.
the camera is great in build quality, image, features But totaly useless

[John_254650199167188]

I have RLC 520 and RLC 522 cameras and am having the same issues as everyone else on this thread (and probably all Reolink customers). Adjustment of the sensitivity does not fix the sun/cloud problem. My cameras are not pointing at direct light, just the patio. Just hoping to bump this back up to the support staff attention, as getting notifications because a cloud keeps passing in front of the sun makes motion detection notifications useless on these cameras.

[joseph_1979]

@john_254650199167188 Detection on these cameras is based on pixels change per frame with the sensitivity giving the depth of changes. So once there is a change then the camera will trigger an alarm. So in order to avoid false alerts another logic is required. And this to determine the object type in the FOV. This is AI and so to get rid of these false alerts you need to upgrade to the A version.

[martinist829_445588003520745]

@joseph_1979 Where's the A version to upgrade to?

[user_718562609025233_718562609025233]

I will soon be installing several Duo 2 WiFi cams one of which will be on a back patio facing East and west patio doors. Hopefully I will be able to define the motion zone so that the patio ceiling fans do not give false alerts..

[ask_670415497359871]

@user_718562609025233_718562609025233 Motion detection can be powerful when enhanced with 3D product animation. This combination improves accuracy, reduces false alarms, and enhances object recognition in various applications, from security to automation and robotics. Don't dismiss its usefulness without exploring this advanced integration.

[richardhennekam68_223400801874070]

Last night we had a thunder storm at around midnight for about one hour. I was getting pre thunder clap notifications by a vibrating watch a couple of seconds before it happened each time. It appears that every time the lightning lit up the sky it was detected as motion. Not good. My cheep camera I used before this one did not do this.

[user_828147483697403_828147483697403]

Almost eight years have past. Has the issue been resolved?