Component Selection Help Needed...

I'm new to the electronics space but having a blast, completed some clever projects but now I have my eyes set on a tougher problem and need some help. This would be just to get a proof of concept together.

I'd be looking to run micro-tensor on a chip that can do basic image filtering.
Those interesting images would be uploaded to the cloud via a cellular modem. I would like Wifi and a low power system that could wake quickly to snap a picture and go back to bed. ESP232?

I would run this all on a lipo rechargeable battery. I can use a solar array to charge the lipo during the day. Math, once the full system is down at min solar output, will get me my lipo.

I need a fairly long-range sensor to detect large objects. HB100 might be a fit?

A camera of 10 MP is probably needed. Images would have to be taken at night as well so I would need to splash a bunch of IR onto the object at a decent distance of 20ft-50ft. I would need a chip to store images of interest while they are uploaded over the cellular chip. There may be changes I'm shooting into the sun so ideas on what to do with that would be interesting. I would take color images when possible but would likely do post-processing in monochrome. Is there a clever way to take a panoramic shot without moving - optics only and stitch everything together on the back end?

My concerns with the cellular chips are the end of life -- all networks get updated so how do I pick a chip that will be around for a while?
I would need a Wifi chip to be able to connect via Wifi from the phone to make sure that the unit is properly set up via stream. I could use it as well to upload images if it's in range.
Would a Zigbee work - deff longer range but the 250kbs connection speed seems tough to work with.

I'd also be interested in the setup has tiled or moved a significant distance -- is there a way to have a gravity trigger give me this info I want to stay away from accelerometers due to the power consumption cost? In a situation where this has moved I'd like to see if the GPS has changed or is changing so I would fire up a GPS chip to grab some stats and ship them out over the cellular connection.

A camera with a visible light lens will suck when used for IR pictures. Compare daylight images with night images taken by a trail camera.

Paul

Hey Chavenor,

Work on breaking that project down into pieces. I believe there are two kinds of problems: easy and hard. An easy problem is one that has a straightforward solution, that takes only a few steps to resolve. Hard problems are a sequence of easy problems. The more you factor easy problems out of your hard problem (which this certainly is), the ... 'easier' it will be to solve.