How do you "apply" a new rule after the fact?

[AlKoch]

Hello,

I have two related questions:

1) In several places I have seen a statement that you can define a Rule that has no Actions to "search" the temporary videos for a new "condition". Since Rules that are defined "ahead of time" are (I believe) applied to the video stream as it comes in I don't see how we can define a Rule "after the fact" and "activate" it so it will generate clips in the Search list. How is this done?

2) I am curious how we would do the following: If we see a segment in the temporary video stream how can we "capture/extract" to keep it? For example, if I see my cat do something funny and I know that the camera is on but I don't have a Rule hat would have captured this, how could I scan the temp video to locate it (I'd know the time) and then mark out the segment to be Exported?

Thank you.

[Devin]

Hi Alkoch,

I'll try my best to answer your questions:

1) When you create a new rule and select it in the Search View, Sighthound will search all temporary video for clips that match. Please see https://www.sighthound.com/support/managing-your-video for more details

2) If you know the time this even happened, select "All objects" in the search view and click on the time bar to the approximate time. Once you have located the video you can go to "Tools > Export clip" to export the video. For details on navigating the search view please see https://www.sighthound.com/support/search-view

Regards,

-Devin

[AlKoch]

Hi Devin,

Your explanation makes perfect sense. Sorry I had to ask that question - don't know why I didn't understand that before I asked!

I do have one clarification I'd like to ask about. I understand that Sighthound is "always" recording "temporary" video and that it applies Rules to that stream to see if it should "mark" out a segment of the recorded video to "label" as a permanent clip. Does "always" really mean always as in "24x7" or does it mean "just when any motion is detected" (under the assumption that it's only something moving that we'd ever care about)? There would be a big difference in disk space requirements depending on which way it works!

Thanks.

[Devin]

Hi Alkoch,

Temporary video is all video (24/7 if you have your rule scheduled 24/7) regardless of motion detected. The person/object detection needs this temporary video so that it can analyze and track all movement between frames and decide what is a person or an object or what should be ignored.

Regards,

-Devin

[AlKoch]

Makes sense - thank you very much for all your help!

[Semper Vaporo]

Maybe how I understand the system will help you (I don't guarantee this is how things are done, but it "works" for me in my getting my fingers dirty fiddling in the system).

I consider that each camera's data is written continuously to the harddrive in its own Daily directory. The data is broken into files, but the "whole" is one contiguous stream of video data. It is broken into files because if there is a power failure ALL the data would be lost because files have to be "closed" before they are "real" on the drive. Breaking it into short (time wise) files limits how much video would be lost on a power failure (or system crash, etc.).

A triggered "Rule" causes time values to be stored in a database that also indicates which file the data starts and ends in (might span two files).

In the "Search" mode, you see a list on the left that is derived from the database. Click on an entry there and it will highlight and that segment is played in the window on the right (might be the end of one of the files and the beginning of another).

Select to view only one Camera's data and click an entry in the List, you can let it complete (or click Stop and then drag the progress marker under the window to the right end), then click Play again and the video will continue into the next part of the video stream on the disk, even if no rule has marked it. The "highlight" on the left may appear as just a thin dark line between two entries.

You are viewing video that is "between" the marked segments, and it might be a whole 24 hours of boring video if there was no rule triggered at all.

You can fast forward through the video with the "+:02" button or drag the progress marker to the right to spot check through the whole thing. When you get to the end, click Play again and you might see more video between Rule marked segments or you my step into the next marked segment.

You can also go backward through the whole video stream by using the "-:02" button and dragging the progress marker to the left.


This may help understanding your other question about "Quick" vs "Permanent" delete. I have made the assumption that "Quick" delete just deletes the entries in the database, but leaves the files on disk, but "Permanent" delete has to remove the data segment from the individual files, thus a large amount of time is spent adjusting files on the disk.

When playing "unmarked" video, right-click on the video play window and you get a dialog to Export or Delete a 2 minute portion of the video. The clip is centered around where the timeline is when you right-clicked (adjusted for if you are within one minute of the beginning or end, you get a whole 2 minutes exported or deleted).


As an aside, IF "you" are willing to accept the responsibility for shooting yourself in the foot, (like I am,) then you CAN delete files and directories in the SightHound structure and SightHound does a remarkable job of recovering from not finding what it expects. I consider that a sign of excellent programming practice on the part of the programmers of VDV/SightHound! I still have all my toes, but I have winged myself a few times in the past playing with data/files/directories that we are all warned to leave alone (and not necessarily has it been SightHound, the Windows Registry is another area where one can have a lot of fun/angst doing things that we are warned are not designed to be altered by the user!... I limped for several days with one of those escapades!).

[AlKoch]

Hi Semper,

WOW! Thank you VERY MUCH for taking the time to write up all of your "notes". This is an incredible wealth of helpful information. I wish I had seen this a week ago before I plunged into this. I agree with you that this is a very nice, well written piece of software (I am a professional developer myself so I pick up on "little things" that can reveal a lot). It's really nice but I think the documentation needs to be expanded a bit (if nothing else, it would probably save a huge amount of tech support time). I have already archived your comments in my notes for Sighthound - it's a treasure trove!

I encourage Sighthound to "license" this info from you and post it some place prominent. It sure answered a lot of questions for me.

Your notes bring up one question you might be able to answer. Your notes mention the +-:02 button. I swear I saw a series of 1x, 2x, 3x buttons on the clip player for adjusting the play speed but I haven't seen them now in a couple of days - and I don't see the +-:02 buttons you mentioned. Am I confused - if not, can you tell me where these buttons are (went)?

Again, your comments are REALLY APPRECIATED. Thank you.

[Semper Vaporo]

I bet you have narrowed the SH window too much. They will come back if you widen the window. I think the 'Speed' buttons (x1, x2, etc.) "disappear" first when changing the window size. Also, there is a "click and drag" area between the "list" and the "view" where you can adjust the relative sizes in the program window that affect how many tools there is room for.

[AlKoch]

Hi Semper,

That was exactly it! Thanks for the answer.

Since you are obviously very familiar with the system I'm curious about a couple of things.
- What cameras are you using? Are any rated for outside use and are you happy with your choice? (I am using Samsung SmartCam HD Pros. At least for the one week I've had them they have been good. Nice quality video but they sure "run hot" and they aren't designed for outside use.
- What computer resources are you running with? I just added a 4th camera and Sighthound started warning I am at 100% CPU utilization. I'm running in a VMware virtual machine with 2 cores and 2GB RAM. Not sure what I'm going to do about this problem yet.

If you feel like sharing this info that's great - if not I understand.

Thanks again very much for you prior post!

[Semper Vaporo]

This is certainly drifting from your original question, but since you asked here, I'll respond here.

Presently my Surveillance system consists of:

A tower PC to better distribute and dissipate heat. It is an HP P7-1010, AMT Athlon ™ II X4 645 processor (quad core) 3.1 GHz, with 6GB of memory and a terabyte harddrive, running Winders 7 (64-bit). It has no internet connection and thus is not running any anti-malware software so all is devoted to SightHound (and my alarm clock program). I do run Task Manager to keep tabs on the CPU usage. It also has some utilities I run periodically to monitor the system or to access the setup screens of the cameras that SightHound does not provide access to.

None of my cameras are mounted outside; they all peer out windows except one that monitors a remote weather station display so I can correlate weather data (wind speed, temperature, etc.) with events on the other cameras.

I have five USB Webcams and one IP (network) cam:

1 East – MS LifeCam Cinema
2 South - HP 3100 Webcam
3 North - Logitec PRO-9000
4 West – TV-IP400W TrendNET – (this is the IP cam)
5 Drive – Logitec Webcam 120
6 Weather – Logitec HD Webcam C270

I have been unable to use more than one of any one model of Webcam… it seems to be a problem with the USB drivers not enumerating individual instances of each camera that is plugged in and the system gets confused as to which camera is which if they have the same name and they will randomly swap views in any video program that tries to access them, even one at a time! And eventually I get a BSoD or the computer will just spontaneously reboot.

With the present system, during the daytime the CPU usage runs between 70% and 80% when nothing much is happening, but will approach 90% when a rule is triggered or when the system seems to be doing some background cleanup.

I want to add one more IP cam that will be mounted outdoors, and maybe others, but I think I am maxed out using Webcams with the present hardware. I am looking for a “cheap” outdoor (IP66) IP cam that will do remote Pan and Tilt, is cheap, uses POE, is cheap, at a decent resolution and frame rate, and is cheap. So far, the “cheap” criteria is the only one that is preventing me from getting one.

If I can provide you with more info about any of this or why I am using what I have, I will be glad to expound on it.