Human / Face detection
- Thread starter 35s77
- Start date
I can't remember if you said you dialed in the camera to your field of view or still running on default settings?
Keep in mind that tripwires work in 3-D and half of the object has to pass thru the tripwire for it to trip, so I can see several instances where a person could walk on the edges of it and not meet the criteria to trigger.
Have you considered intrusion zones instead of IVS? Most of us test it with both and then go with the one that is more reliable. Most of mine intrusion box works the best.
Many of us have also noticed the motion triggering of the 180 isn't as good as the other Dahua offerings. But I would think it should work fairly well in your field of view as it is confined to your yard. But many have noticed outside of the center third of the field of view it can struggle in certain situations.
I did dial in the night time exposure settings, but left daytime as default.
ok yes it for sure struggles outside the center third most. But I just had my daughter walk straight down the sidewalk and side of driveway/yard thats in the center of those zig zags, left to right then right to left and it was 50/50 on picking her up, even in the center third.
I had thought I read the tripwires were more reliable than the intrusion - I will try the intrusion boxes and see if it makes a difference. Thank you.
ok yes it for sure struggles outside the center third most. But I just had my daughter walk straight down the sidewalk and side of driveway/yard thats in the center of those zig zags, left to right then right to left and it was 50/50 on picking her up, even in the center third.
I had thought I read the tripwires were more reliable than the intrusion - I will try the intrusion boxes and see if it makes a difference. Thank you.
OK I didn't do it that way, but there is an asterisk on the clones. should I check the clone box on the main cam or redo the clones?
Mike A.
Known around here
- Joined
- May 6, 2017
- Messages
- 3,950
- Reaction score
- 6,637
I've not seen that tripwires are any more reliable. I do see that where I have both overlapping, generally the intrusion box will be what triggers first. But that could just be my views and how I have things set up. They both have their place.
Mike A.
Known around here
- Joined
- May 6, 2017
- Messages
- 3,950
- Reaction score
- 6,637
Yeah, check clone master for the first. If it's showing the asterisk, then it should be OK but look at the camera status page. Do you see a bitrate for the clone cams? If not, then good. If so, then they're just running as another cam. BI does some odd things with clones/copied cams sometimes.
How are you doing the different rules by cam? Copying the one with all of them set up and then disabling those you don't use on the clones?
ok i checked the clone box on main camera. no bitrate on the clones. i redid the IVS to be all intrusion boxes. I am goin through now and find/inspect cameras.
Mike A.
Known around here
- Joined
- May 6, 2017
- Messages
- 3,950
- Reaction score
- 6,637
Good idea to redo them I think. They should work as spotters either way but not running as a "clone" will cost you a lot more in terms of resources. And you'll have N devices pulling streams from the cam vs one so might have something to do with the lost signals.
ok everything redone, new clones, new intrusion boxes instead of tripwires. I will let it run a while. I do notice the intrusion boxes are triggering for cars going by in street while intrusion box is short of the street so guess i need to pull those in a little more off the street.
no unexpected triggers, and the intrusion boxes seem to be much more accurate. Cars on the street are triggering the boxes though, and its no getting some humans on the edge of the boxes until they cross like halfway in. so ill have to play with it and figure out how to stop getting cars in the street to trigger and start getting humans on the edge. maybe i extend the boxes into the street and say only human.
Mike A.
Known around here
- Joined
- May 6, 2017
- Messages
- 3,950
- Reaction score
- 6,637
Check both appear and cross and set direction to both for the boxes. Unless you have some reason not to.
That is one advantage to the tripwires. When zig-zagged as usual you have a lot of potential points where things will trigger again if it misses an initial crossing as for the intrusion box with only cross selected. But the "appears" should and usually does pick those up well.
Still seeing the ONVIF errors with the clones re-done?
That is one advantage to the tripwires. When zig-zagged as usual you have a lot of potential points where things will trigger again if it misses an initial crossing as for the intrusion box with only cross selected. But the "appears" should and usually does pick those up well.
Still seeing the ONVIF errors with the clones re-done?
yes i have them set to both directions and appear/cross. I made the boxes taller into the other side of the street and switched them to be human only. The triggers in the spotter cameras seem to be working great now. I'll figure out cars in the driveway later if I want. Guess I could just make another set of intrusion boxes well into the driveway for cars only and list those rules in the ONVIF configuration of each spotter.
Now a person is well inside the box if they are stepping foot in my yard and the appear seems to be picking them up.
One of the cams did get a new error: PTZ:12002: /onvif/ptz_service error but hopefully something I just did wrong. The other errors seem to have stopped. They were there when coming through when I first set the cameras up but haven't gotten one in over an hour.
Hard to tell what is going on. When I watch people move through the IVS rules I have set up in the main 180 cam, the camera that has all the IVS rules for the PTZ - its triggering almost 100%. But the PTZ isn't acting right.
Still getting the subscription errors on main/clone spotter 180deg cams. Now I am getting a new one. Getting these:
Events: subscription 8000ffff
Events: subscription 00002ee2
And it seems like - when the PTZ is tracking someone - it will get triggered by another cam and pan to that preset it was tripped on but seems like it's happening late. Like it should not be pulling off something it is tracking, and then when it pans to where it was tripped, the person is no longer there.
So if I have PTZ presets set up 1, 2, 3, 4, 5 left to right, and IVS rules in the spotters in roughly the same spots. The PTZ will be tracking someone walking left to right (1 to 5) and around 3 it will trigger back to 2 but the people are gone...they are in 3. If that makes sense.
And then other times the PTZ will randomly shoot up into the eve in the middle of tracking, where I have no preset or IVS rule.
So it seems like maybe the spotting cam/clones are losing connection - maybe fast enough where its no registering in the log? I wonder if it's the camera, the wire, or a setting.
But I thought the PTZ should track, and when it stops tracking go back to its home preset, and only when it was back in its home spot could it be triggered again. So I don't know.
So the PTZ wondering off to the eave is probably it losing the tracking object "signature" and a separate issue. That can be confirmed if you have an SD card in the PTZ and playback the event with IVS rules on and see where the box goes.
Hmm every one of those example videos you just posted could either be the PTZ losing the object "signature" and lost track or the preset is being called.
I think if you don't have an SD card in the PTZ, now would be the time to add it because that will let you know whether it is a PTZ tracking issue or a BI spotter call issue.
Hmm every one of those example videos you just posted could either be the PTZ losing the object "signature" and lost track or the preset is being called.
I think if you don't have an SD card in the PTZ, now would be the time to add it because that will let you know whether it is a PTZ tracking issue or a BI spotter call issue.