Security camera/CCTV minefield.

[ajlelectronics]

Would be intrigued to see the outcome of this, very clear picture of the vehicle and the driver. I would be really disappointing if they fob you off with some "Not enough evidence" story

Sadly many businesses/home leave this kind of thing out to be collected and it justs encourages people to be around.

Oh I had much more than that. I contacted the Police and they did their best to ignore me, but persistence paid off and I was visited by a really helpful and professional PC who took the time to come to my home address for a statement. They know who it is and where he lives so was going to have a word immediately after. I gather that the perp. has been "invited" in for interview but I will hear nothing until that interview or an arrest has been concluded.

When it is all over, I can perhaps get away with posting the video but it may be sub justice at the moment.

 

[brewdexta]

I've been looking to replace the Hikvision and Sannce DVRs with something better. I'm pretty sure they come from the same stable software wise, not very good in my opinion, especially the need for ancient end of life flash plug-ins for Internet Explorer to view over the web. I also wanted to cut power consumption the bone.

The missus wants more wildlife cameras for Christmas so took that as an ideal opportunity to try something else. I've had Blue Iris and similar running in the past, but they can be a CPU hogs and didn't want to use a sizeable power hungry PC for it. I have an rpi4, but they tend to be a bit weak for any more than one or two cameras. Then my boffin mate told me about Frigate and the Google Coral accelerator. This can run on a pi and hands off object detection to the accelerator that can perform 4 trillion operations per seconds with only a few watts, I thinks its 2 watt per trillion ops Motion detection is still done in CPU but its more lightweight due to the object detection, and supposedly, you get less false positives too.

So not only motion detection, but object detection, so it will tell you if a bird has flown into the picture, and what type of bird. Here's a list of the main functions.

Tight integration with Home Assistant via a custom component
Designed to minimize resource use and maximize performance by only looking for objects when and where it is necessary
Leverages multiprocessing heavily with an emphasis on realtime over processing every frame
Uses a very low overhead motion detection to determine where to run object detection
Object detection with TensorFlow runs in separate processes for maximum FPS
Communicates over MQTT for easy integration into other systems
Recording with retention based on detected objects
Re-streaming via RTMP to reduce the number of connections to your camera
A dynamic combined camera view of all tracked cameras.

The accelerator is normally about 80 quid, but due to demand its currently up about 120 quid in places its in stock, I already have the pi 4, a PoE switch and a decent Hikvision 4MP camera to play with so will get it all working and see if its any good, If it is, I can move the other cameras later.

So, I have a new geeky project

Here's vid of it being used with Home Assistant and how to install it. There is a bug in the latest Linux kernel that stops the accelerator working, so I'll just use an older kernel until its fixed.