Hi,
I'm developing WildCamera, a new open source wildlife camera system that makes use of the very strong MIPI CSI-2, ISP, hardware video encoder and NPU in the i.MX 8M Plus:
https://github.com/William-Robert-Robertson/WildCamera
A requirement from our conservationists and ecologists is that they would like the system to be able to start video recording within c. 200 milliseconds of an animal being detected. We work with Europe's most endangered dormouse, squirrel and bat species and these species can move very fast so speed is important.
To achieve this, we plan to keep a cut-down version of embedded Linux on the i.MX 8M Plus in a low power sleep with RAM in self-refresh and have a simple digital signal sent to waken the system when an animal is detected.
To make the process of wakening from sleep state as fast as possible, our cut-down version of Linux would not include time-intensive functionality like Ethernet, WiFi, USB, graphical user interface, output to a display, keyboard or mouse interaction, etc. Early indications are that our cut-down version of Linux will only require small amounts of RAM to run - less than 2GB would be no problem.
Linux would have to wake up to the point of being able to receive video data from the MIPI CSI-2 interface to a camera module, processing it through the ISP and hardware video encoder on the i.MX 8M Plus and writing it to SD card storage.
Is it practical for us to expect to be able to wake Linux from a low power sleep state to the state of being able to accept video data from a MIPI CSI-2 image sensor within of the order of 200 mS?
Should we consider possibly temporarily buffering video in RAM before writing it to the SD card to speed up the wake-up process or would this not be neccessarry?
The system will be solar powered via a battery and independent in the field so energy efficiency is important - we hope that keeping Linux on the i.MX 8M Plus in a low power sleep mode and wakening it only when an animal is detected approaching will achieve a good balance of speed and energy efficiency.
For dormouse species we rejected the possibility of booting embedded Linux when and animal is detected because measurements indicated that reducing the boot time of Linux below 2 seconds would become extremely difficult - though we're keeping this option open for slow moving animals. Please let em know if my reasoning on this might not have been valid.
Some more information about our conservation work is here:
https://new-homes-for-old-friends.cairnwater.com/
Thank you very much for your help!
Will
