Hi! Could you recommend a free software for creating a local video surveillance archive with IP cameras under Windows? I need video recording based on motion detection (only when a person appears in the frame) and continuous time-lapse recording 24/7.
Video Surveillance Software for Windows
- Thread starter websecurity
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
This is probably the most common windows based software for smaller setups(under 24 cams)
blueirissoftware.com
Home - Blue Iris Software
Innovative Video Management Software (VMS) to record up to 64 cameras, both IP and analog. Keep an eye on your home, place of business & valuables.
blueirissoftware.com
Some Reolink cameras can do it for you and you only need an ftp server for them to upload the videos (or you use an SD card in the cameras).
There's not much point in streaming all the video from all the cameras to a server that then tries to detect stuff like motion or persons by examining all the streams and gets overwhelmed with so many cameras. (It actually sucks to do that.)
For time lapse, I guess you'd just let the server peridically grab the current image from each camera. Or perhaps you can set an email alert; I'm not sure what the email timer is for:
There's not much point in streaming all the video from all the cameras to a server that then tries to detect stuff like motion or persons by examining all the streams and gets overwhelmed with so many cameras. (It actually sucks to do that.)
For time lapse, I guess you'd just let the server peridically grab the current image from each camera. Or perhaps you can set an email alert; I'm not sure what the email timer is for:
Attachments
It could work, but heavily depends on the camera hardware, some of cheap ones will detect leaves/shadows moving as "human" no matter what thresholds you try to get, and others can reliability detect humans, cars etc.. and even do tracking.
Examples of the later category:
The Reolink camera can even detect the automatic lawn mower as an animal 
Detection with those seems to work pretty well; I haven't even altered any thresholds. That kind of thing really needs to be done in the camera and not on a remote recording device.
BTW, the timer apparently would send a picture every 5 minutes during alarm hours. Maybe they'll improve on that in future firmwares since this isn't exactly flexible. But if you want a timelapse with like a picture per day, it would kinda work and you can delete the other 5 pictures you'd get.
Detection with those seems to work pretty well; I haven't even altered any thresholds. That kind of thing really needs to be done in the camera and not on a remote recording device.
BTW, the timer apparently would send a picture every 5 minutes during alarm hours. Maybe they'll improve on that in future firmwares since this isn't exactly flexible. But if you want a timelapse with like a picture per day, it would kinda work and you can delete the other 5 pictures you'd get.
yes, and Blueiris supports up to 64 cameras per instance.
other solutions like Agent DVR are quite similar.
plenty of options:
Zoneminder
Frigate
AgentDVR
Channels DVR
Scrypted
Shinobi NVR
Zoneminder
Frigate
AgentDVR
Channels DVR
Scrypted
Shinobi NVR
I checked out Blue Iris, and it’s pretty clunky. The interface is heavy, cluttered, and feels a little like it’s from the 90s. There’s no time-lapse feature or built-in object detection either.
yep, the Blueiris UI is "outdated". But i hope they keep it that way since "customers" don´t like changes in the UI. The last major UI changes are 10 years old, most of the solutions mentioned above have not even been around that long.
Time-lapse => you can just define it like that to get a picture all 10 seconds (like in the follow screenshot) and depending on your needs run ffmpeg every hour/day/week to combine it. A time-lapse for an eventbased trigger is not the best idea, you may not get the full "picture" what happend. Today you typical just record a continuous "substream" to keep the storage usage low.
Object-detection: Blueiris should just bring you a "popup" if you would like to install the AI stack (code project). But you can also use different solutions like the good old deepstack or blue candle, both are quite easy to integrate.
If you don´t like the Blueiris UI, have a look into AgentDVR (iSpy - Best Video Surveillance and Security Camera Software for Motion Detection and Remote Monitoring). There is also a "free" Version of Milestone, but thats a quite heavy solution (XProtect® Essential+ | Free VMS from Milestone Systems).
Time-lapse => you can just define it like that to get a picture all 10 seconds (like in the follow screenshot) and depending on your needs run ffmpeg every hour/day/week to combine it. A time-lapse for an eventbased trigger is not the best idea, you may not get the full "picture" what happend. Today you typical just record a continuous "substream" to keep the storage usage low.
Object-detection: Blueiris should just bring you a "popup" if you would like to install the AI stack (code project). But you can also use different solutions like the good old deepstack or blue candle, both are quite easy to integrate.
If you don´t like the Blueiris UI, have a look into AgentDVR (iSpy - Best Video Surveillance and Security Camera Software for Motion Detection and Remote Monitoring). There is also a "free" Version of Milestone, but thats a quite heavy solution (XProtect® Essential+ | Free VMS from Milestone Systems).
The features you mentioned are all available in SmartVision Dev (object detection, face recognition, time-lapse recording, etc.). A big advantage is that it uses fewer resources compared to similar software.
on the first view some nice features like includes models for smoke & fire detection and a tool to train new faces. The object dectections seams to require a subscription or a "free" cloud account.
I also can´t say that it using fewer ressources. In the end it´s just a very very basic gui for a collection of various opensource tools (vlc, tensorflow, ffmpeg, opencv,...) and some opensource(?) models for object recognition. The gpu support is limited to the outdated librarys from vlc/ffmpep => I would say it is comparatively resource hungry.
I could not really find a correct indication of the corresponding licenses of the used opensource software in the tool or on their website. This is the kind of behavior that makes software not really recommendable for me.
