Thanks folks! I’m going to try out those options. I like the idea of being able to define an area, that way dumb minor movement like shadows etc..could be excluded from being compared.
I’m just looking to record a human delivering a package. My address is cursed (long story) so packages are often marked as delivered but they are somewhere else. Same with food delivery orders! So I want a timestamped video showing LIES!!![:)]( "Smile")
I would even be OK with having the camera record nonstop or in 30min chunks or a certain file size chunk and then start a new one. My car dashcam works that way.
Is it possible to record say daylight (6AM-9PM) hours and then stop and start again the next day? The only rub there is disk space. How much space would I need to record at a medium quality rate where you can make out people? I have a 500GB NVME installed currently with an identical one via USB that I use to make a backup of my computer before major changes (almost like a snapshot). So that could be another 500GB on the table.
Worst case I buy larger NVMEs. Any clue on space usage? Surely I’m not the first to try and do continuous recording.
Oh I’m also looking at a project called “Motion” that looks promising on paper. I dunno I need help. I went as far as to order a microwave radar sensor capable of a decent distance to maybe use that as a cue to record. Sensor = significant movement—> record for 1 minute then stop.
Open to all kinds of ideas. Sadly I’ve never worked with vision coding so I’m basically just reading the RPi manual. I understand the concept of frame differencing but how it works in code line by line is lost on me. Hence I’m open to just recording nonstop if I can make it fit. It’ll be deleted that same day by me or a script to ensure the next day it doesn’t crash the drive by filling it!
How would you do it?
Get the AI Camera? Is that really needed for such a simple task?
Thanks in advance for your support!!!
I’m just looking to record a human delivering a package. My address is cursed (long story) so packages are often marked as delivered but they are somewhere else. Same with food delivery orders! So I want a timestamped video showing LIES!!
I would even be OK with having the camera record nonstop or in 30min chunks or a certain file size chunk and then start a new one. My car dashcam works that way.
Is it possible to record say daylight (6AM-9PM) hours and then stop and start again the next day? The only rub there is disk space. How much space would I need to record at a medium quality rate where you can make out people? I have a 500GB NVME installed currently with an identical one via USB that I use to make a backup of my computer before major changes (almost like a snapshot). So that could be another 500GB on the table.
Worst case I buy larger NVMEs. Any clue on space usage? Surely I’m not the first to try and do continuous recording.
Oh I’m also looking at a project called “Motion” that looks promising on paper. I dunno I need help. I went as far as to order a microwave radar sensor capable of a decent distance to maybe use that as a cue to record. Sensor = significant movement—> record for 1 minute then stop.
Open to all kinds of ideas. Sadly I’ve never worked with vision coding so I’m basically just reading the RPi manual. I understand the concept of frame differencing but how it works in code line by line is lost on me. Hence I’m open to just recording nonstop if I can make it fit. It’ll be deleted that same day by me or a script to ensure the next day it doesn’t crash the drive by filling it!
How would you do it?
Get the AI Camera? Is that really needed for such a simple task?
Thanks in advance for your support!!!
Statistics: Posted by anglerfish27 — Fri Dec 13, 2024 4:49 pm