Sentry Person and loitering recognition, Vehicle detection & identification, + Price increase

And you know that because you have been here with me when I was testing, come on!

If I trigger the camera and then enter the frame a second after then I am missed, so if I send a football in then I walk in the frame I get no sentry notification. I have tested this extensively to find when alerts would be missed. Hence why I have been looking at alternatives.
I know that because you posted nonsense that doesnt prove how many images are sent. Repeatedly. You didnt believe me when I told you they already stated that they take more than one image, so I posted the link for you.
 
And you know that because you have been here with me when I was testing, come on!

If I trigger the camera and then enter the frame a second after then I am missed, so if I send a football in then I walk in the frame I get no sentry notification. I have tested this extensively to find when alerts would be missed. Hence why I have been looking at alternatives.

This post is from a month ago or so, maybe the very latest version of BI is doing this. But it is still not good enough, I would want to have it take the messages that I would get in my email, say one every second for 3 seconds or maybe for 5 seconds. Maybe even two a second.

I suggested all this to them, they acknowledged the points I made but I never heard a definitive yes or no and when this would happen.
 
This post is from a month ago or so, maybe the very latest version of BI is doing this. But it is still not good enough, I would want to have it take the messages that I would get in my email, say one every second for 3 seconds or maybe for 5 seconds. Maybe even two a second.

I suggested all this to them, they acknowledged the points I made but I never heard a definitive yes or no and when this would happen.
The post is 4 days shy of being two months old. Irrelevant as your post was today. I never said it cannot be improved. My post is your tests dont have any bearing on the number of frames sent.
 
This post is from a month ago or so, maybe the very latest version of BI is doing this. But it is still not good enough, I would want to have it take the messages that I would get in my email, say one every second for 3 seconds or maybe for 5 seconds. Maybe even two a second.

I suggested all this to them, they acknowledged the points I made but I never heard a definitive yes or no and when this would happen.

Just because they stated it doesn't mean they are in fact doing it right now.

The point here isn't if am right or if you are right about this. The key point here is that Sentry alerts can be easily circumvented and alerts cab be easily missed and this my friend I have tested extensively and have reported to them as well.
 
Just because they stated it doesn't mean they are in fact doing it right now.

The point here isn't if am right or if you are right about this. The key point here is that Sentry alerts can be easily circumvented and alerts cab be easily missed and this my friend I have tested extensively and have reported to them as well.
They had been saying that they will be implementing it for many months. They finally did. Now you are suggesting they stopped doing it?
You cannot be right. Even if sentry says that only one clip is sent you are still wrong. All I said was that nothing you posted evidences that only a single image has been sent. Yet you keep going in circles.
No one said its perfect. Its a work in progress. Perhaps you need a live monitoring service to protect your fort knox.
 
The post is 4 days shy of being two months old. Irrelevant as your post was today. I never said it cannot be improved. My post is your tests dont have any bearing on the number of frames sent.

There is one more important more, the post you linked here talks about more images for accuracy. While accuracy is very important, if your coverage is only a second then it doesn't matter how accurate your detection is. My idea to send an image to Rekognition every few seconds would solve this problem, I have also suggested this to sentry but I don't think they were very keen on it.

The ideal solution would be to have the video analysed in real time to find people and/or anything else of interest, while this is possible it is costly and not financially viable for most. That company Sentinel Black was doing it but they seem to have vanished.
 
They had been saying that they will be implementing it for many months. They finally did. Now you are suggesting they stopped doing it?
You cannot be right. Even if sentry says that only one clip is sent you are still wrong. All I said was that nothing you posted evidences that only a single image has been sent. Yet you keep going in circles.
No one said its perfect. Its a work in progress. Perhaps you need a live monitoring service to protect your fort knox.

Thanks, looks like only you can be right.
 
There is one more important more, the post you linked here talks about more images for accuracy. While accuracy is very important, if your coverage is only a second then it doesn't matter how accurate your detection is. My idea to send an image to Rekognition every few seconds would solve this problem, I have also suggested this to sentry but I don't think they were very keen on it.

The ideal solution would be to have the video analysed in real time to find people and/or anything else of interest, while this is possible it is costly and not financially viable for most. That company Sentinel Black was doing it but they seem to have vanished.
keep going in circles. Rekognition is an amazon product of course sentry would not want to sent an image there.
Sighthound software has built in local AI.
 
keep going in circles. Rekognition is an amazon product of course sentry would not want to sent an image there.
Sighthound software has built in local AI.

You keep taking what I say out of the context, I never said sentry should do this if they did they wouldn't exist.

I said to you in another reply BI should, I can, other tools can.... keep the context please. I was suggesting alternatives. I am trying to solve a problem here so please don't be sarcastic
 
You keep taking what I say out of the context, I never said sentry should do this if they did they wouldn't exist.

I said to you in another post BI should, I can, other tools can.... keep the context please. I was suggesting alternatives. I am trying to solve a problem here so please don't be sarcastic
Wait, so you can be sarcastic and I cant? Oh how quickly you forgot about post 60

Everything is very easy when you dont know what it takes or costs to do it. Perhaps you give it a go.
 
Wait, so you can be sarcastic and I cant? Oh how quickly you forgot about post 60

Everything is very easy when you dont know what it takes or costs to do it. Perhaps you give it a go.

I actually do know how much it would cost to put some integration together with rekognition and how much it would cost to use the rekognition as a service too. I have done the proof of concept already.

I just dont want to do and maintain it myself, I want to pay for someone's service hence I am using Sentry for now.
 
I actually do know how much it would cost to put some integration together with rekognition and how much it would cost to use the rekognition as a service too. I have done the proof of concept already.

I just dont want to do and maintain it myself, I want to pay for someone's service hence I am using Sentry for now.
You know nothing. IF you could your would have already done it. You havent even done the basic cost analysis of using rekognition. You suggested an image be sent every few seconds (which is sill not enough to eliminate the issue if a person can pass the scene in under 3 seconds). Take your average camera how many minuets of recorded triggers do you get on average per day? Lets be conservative at 30 minuets. That is 1800 seconds, 1800/3=600. 600*.001=0.60. You are now at 60 cents per day per camera at a relatively low 30 minuet trigger time.
 
You know nothing. IF you could your would have already done it. You havent even done the basic cost analysis of using rekognition. You suggested an image be sent every few seconds (which is sill not enough to eliminate the issue if a person can pass the scene in under 3 seconds). Take your average camera how many minuets of recorded triggers do you get on average per day? Lets be conservative at 30 minuets. That is 1800 seconds, 1800/3=600. 600*.001=0.60. You are now at 60 cents per day per camera at a relatively low 30 minuet trigger time.

Ok, you know better.
 
Hey guys. Just jumped into IPCAMTALK today. Just to cool the flames, right now Sentry is still integrated to take only one frame per motion clip. Apologies if there was some misunderstanding from earlier/past posts. We understand all the limitations mentioned by VinAU regarding this first go at a solution and also appreciate the feedback provided to support to make the product better. We are working on a more accurate solution but it will probably be offered at a different price point for the economics to work. Still TBD on next steps.

-- Sam from Sentry
 
  • Like
Reactions: VinAU
Hey guys. Just jumped into IPCAMTALK today. Just to cool the flames, right now Sentry is still integrated to take only one frame per motion clip. Apologies if there was some misunderstanding from earlier/past posts. We understand all the limitations mentioned by VinAU regarding this first go at a solution and also appreciate the feedback provided to support to make the product better. We are working on a more accurate solution but it will probably be offered at a different price point for the economics to work. Still TBD on next steps.

-- Sam from Sentry
How do you explain this contradictory post? Sentry Server Error 400 with newest BI 5.1.0.3
 
  • Like
Reactions: joshwah
How do you explain this contradictory post? Sentry Server Error 400 with newest BI 5.1.0.3

Thanks Sam.

I have put a strategy together to use with Amazon Rekognition at a price point that I would be happy to pay and some conditional rules that would prevent or reduce the images being sent if a known person is found in the frame. Happy to discuss.

Sentry has been very important in a number of cases I worked with the local police so it has provided good value but there are still issues with the approach as I mentioned before and I do recognise that even with multiple images sent at the beginning, it wouldn't solve my problem so I need a more complex approach.