Review-OEM 4mp AI Cam IPC-T5442TM-AS Starlight+
- Thread starter looney2ns
- Start date
sebastiantombs
Known around here
I've been running the 5442 using both BI motion detection and IVS, ONVIF triggers. I'm plagued with false triggers due to shadows with both, especially in the morning. The camera looks west and there's a big spruce just east of it that waves, so nicely, in a gentle breeze. So, yesterday I switched it to facial recognition since it really covers the approach to the front door area. That got rid of the false triggers but last night it didn't work out so well.
When in facial recognition mode control of the image, gain, compensation and so on, all can't be adjusted. I keep the camera in color mode 24/7 and without being able to control things color at night just doesn't work. The firmware is V2.800.0000000.11.R, Build Date 2019-12-30, which I think is the latest. Anyone else notice this? Is there a reason why normal "tuning" adjustments can't be made in facial recognition mode?
When in facial recognition mode control of the image, gain, compensation and so on, all can't be adjusted. I keep the camera in color mode 24/7 and without being able to control things color at night just doesn't work. The firmware is V2.800.0000000.11.R, Build Date 2019-12-30, which I think is the latest. Anyone else notice this? Is there a reason why normal "tuning" adjustments can't be made in facial recognition mode?
bigredfish
Known around here
I still have a lot of false alerts with that same FW using IVS regardless of using the Human/Vehicle filteing also.
bigredfish
Known around here
Headlights passing in parallel, tree shadows , remember mine are the bullet version...? Ever since applied the red triangle fix.......
I may try another factory reset and try again.
Schnoidz
Getting the hang of it
I see now that Amazon has the 2.8mm back in stock.
As an Amazon Associate IPCamTalk earns from qualifying purchases.
If you are getting a lot of false motion events due to moving shadows, you can try this trick...
Create three zones in BI.
Zone A is the entire field of view
Zone B is the edges of the camera's field of view - not too big however because you don't want to create a zone where someone can walk around the edge and not trigger the motion
Zone C is the part inside of zone B. Because you have zone A, you don't have to worry about touching zone B with C, but the two should be close, but not overlapping.
Then create your motion trigger when B>C. This will catch anyone coming into the frame, but will get rid of most false triggers due to shadows. At least this has worked well on my cameras that tend to have a lot of shadow movement.
You don't have to use these zones all the time either. You can create profiles based on the time of day and change the triggers for each profile. So use this profile in the morning when the shadows are bad, and go back to a more normal zone/trigger set up for night and afternoon when the shadows aren't bad.
bigredfish
Known around here
Thanks but not using BI, Dahua XVR.
I did another factory rest today, seems to be working at the moment...
I did another factory rest today, seems to be working at the moment...
sebastiantombs
Known around here
Been there, done that. Problem is the amount of shadow and how it progresses and otherwise moves with the wind.
IAmATeaf
Known around here
I’ve also tried doing this but for me BI would then miss some events. So for example when we weren’t all locked in my kids leave for work at the same time every day and randomly it would not record one of them leaving or arriving back home.
There are no perfect motion detection schemes unfortunately and they're are compromises with every solution.
The theory works because it is very unlikely (ie nearly impossible) for a person to suddenly appear in the middle of the field of view (C zone) without first walking into the point of view from the side (B zone). (Of course there are always exceptions to every rule and your physical location may allow a person to jump into the center of the field of view).
As with any motion detection, the effectiveness is in the details - the min object size, contest, duration, object detection settings, etc etc etc. If you miss events, it may be because the settings are too broad. Of course making the settings tighter may result in more false triggers too.
I just wanted to throw it out there because it's not a solution that people would necessarily think of first - especially people just starting out using BlueIris. For the more seasoned user this is old news, but you never know who is reading the thread.
Again, I'm not suggesting that it is a perfect solution for everyone, but hopefully it helps people think outside the box.
Thanks!
DWinters1979
Getting the hang of it
Hi there, is the 2.8mm better than the 3.6?
Need it to monitor my boat an outboard engine.
Thanks!
tibimakai
Known around here
DWinters1979
Getting the hang of it
tibimakai
Known around here
I'm not an expert, but I believe that you would recognize better with 3.6MM.
The wider the view, the closer someone has to be to be ID'd. So the 2.8 would need someone closer that the 3.6mm for an ID of the face.
bigredfish
Known around here
As the lens size (2.8mm, 3.6mm for example) gets bigger, you see things closer (think zoom) but with a smaller field of view (FOV)
FOV
2.8mm H: 113°, V: 60°
3.6mm H: 89°, V: 48°
Dahua calls it the DORIS distance
FOV
2.8mm H: 113°, V: 60°
3.6mm H: 89°, V: 48°
Dahua calls it the DORIS distance
| Lens | Detect | Observe | Recognize | Identify | |
| 2.8mm | 58m(190ft) | 23m(75ft) | 12m(39ft) | 6m(20ft) | |
| 3.6mm | 72m(236ft) | 29m(95ft) | 14m(46ft) | 7m(23ft) | |
| 6.0mm | 113m(371ft) | 45m(148ft) | 23m(75ft) | 11m(36ft) |