
@GentlePumpkin Any Tips? Regrettably "log almost everything" isn't really helping me Substantially hoping to find out what the issue here is.
Not currently - that AI Port can only pair with one particular digicam today. You will discover studies that it will pair with 3 4K cameras within a potential firmware update, which can be described as a killer element, but for now it is just one camera for every AI Port
I'm unsure of a method to increase a blank camera but you could potentially duplicate just one of one's cameras, increase the weather conditions data and not file the new digital camera. I will see what else I think of
On top of that SenseAI comes along with each of the bells and whistles enabled, none of that happen to be specially crucial and even essential, in a standard movie surveillance circumstance.
I feel that results precision of detection really seriously sometimes, Particularly less than terrible distinction/brightness/lighting cases. For my reasons, eliminating Phony triggers, DS is carrying out an outstanding position so if it ain't broke I'm not fixing it, nevertheless.
v1.forty three Massive up grade: introducing person interface, functioning in history, studies etcetera; however has bug b4 and b5
It will most likely not function for non reflective plates during the lpr evening vision method except possibly very near
tantrim stated: What other choices to acquiring mobile notifications is there? I'm just beginning to Establish/put in my security procedure so I'm fun ai tools nonetheless Understanding. Click to extend...
Maybe SenseAI will get to in which Deepstack was in the event the swap was produced, but I felt for almost any NOOB, finding Deepstack going was relatively painless.
Not one of the objects are in my concentrate on record but do bring about the TPU being 50% much more utilized. Once again possible something which may be reduced with qualified utilization of BI choices like static item detection so not a coral AI situation straight.
The mask image ought to contain the identical identify as the profile file for the chosen digicam. So When the profile file is 'Garage indoor.txt', the mask file for this digital camera needs to be named 'Garage indoor.png'.
To shut All those features down the sole way to get it done is always to edit a .json file. I don't have any difficulty carrying out that, but too many end users do. The additional loading in CPU and memory is just a major squander of CPU, memory and maybe GPU Subsequently.
The chart "Frequency of inform end result confidences demonstrates how frequently Aitool is how self-assured about applicable detections (green) and irrelevant detections (orange). This might assist you to setting a self-assurance threshold that cuts out Fake positives and isn't going to conceal appropriate alerts from you.
sebastiantombs claimed: @kaltertod If you use sub streams in your cameras it is possible to shut of components acceleration on the GPU and Reduce CPU utilization substantially. That will likely free up GPU memory for AI processing. Click to increase...