Leaderboard

+1 for the DND references. (Fellow nerd).

I purposefully avoided ML/AI talk, as too nerdy, LOL. For those who aren't familiar, ML = Machine Learning and AI = Artificial Intelligence. For the point of this thread, they're the same thing. Think of AI or machine learning as teaching a computer program something and asking it to compare new things to what you just taught it. Like when you do a CAPTCHA you've had to solve to get access to a bank or some website. Pick the pictures that contain a stop sign. Pick the pictures that contain a taxi cab. They show a 3x3 block of photos and you click on those that contain the sign. You did the machine learning part when you learned what a stop sign looks like, and you applied the machine learning part when you extrapolated that this red thing in the background of this photo with the letters STO visible is a stop sign, even though the P is not visible. When a computer does this, that's machine learning. You teach the computer by feeding it a directory full of thousands of photos of street corners or stop signs in the dirt or cartoon drawings of stop signs, and after a while, the machine can pick out a stop sign. ML and AI are best at photo and text analysis. In my line of work, in a monitoring department in IT operations in a large corporation, we're tasked with using ML to identify problem areas in logs and transaction data and identify when something could go south soon. We use machine learning to determine who might have an issue, and to predict which way to route certain transactions to meet the most likely success. i.e. you start to see a certain pattern in the data that indicates "not right". We can't see the pattern because we don't even know what the pattern is or could be. All we know is a certain time range was just fine, so we use that time range to train the system on a known-good period, and then turn the AI loose on the logs to determine what no-good looks like and please, computer, just tell us when things look off. And based on the entries, the next time a certain huge transaction comes along, it'll be best to send it to the server with the most available resources, as opposed to the server that is performing fastest at this moment. How does this relate to the furry friends? I see a project in the future to train a machine what humans in the woods looks like with thousands of photos of people in the woods in all conditions, and then overfly a hot spot taking thousands of pictures from all sorts of angles, and then turn it over to the AI to crunch the numbers to see if there are any people or beings in the photos. It could even outline what it thinks is a person to make human review faster. You could teach it deer, squirrel, elk, bear and any number of animals, and have it draw out those frames from video or stills. You could use thermal images too, simply to identify hot spots for analysis. At the very least, you would know where the elk are. This eliminates some of the factors that limited success to this point. It eliminates our lack of mobility and their enhanced mobility by making observations from greater distances, it eliminates rarity of observers, by making each drone an observer capable of thousands of tireless observations a minute. Obviously it doesn't eliminate rarity of target subjects, it's just a compensating control, more surface area = higher chance of encounter. What would this cost? The AI or machine learning toolkits are free. Computers are cheaper than ever, but churning through millions of frames or tens of thousands of stills is going to take compute resources and/or time. A modern gaming PC with a good video card could process thousands of photos an hour. A drone with thermal imaging? That's not going to be cheap. if I had a few tens of thousands of bucks laying around, this is how I would invest it. I would then sell the resulting images of any location that doesn't produce fruit to hunters to increase chances of seeing elk or deer or bear to offset some of the cost. Who knows, maybe that alone is enough of a market to pay for it and your side job becomes bigfoot hunting while your full time job is thermal imaging and machine analysis. Why not?

Now there's another good book for you!

Man made fleece dries out faster. But wool retains its heat retention even when wet. I’ve never jumped in a lake with my filson. But I’ve been in some bad snow storms. Knocking off several inches on the brim of my cowboy hat as I rode on. The red plaid was white with wet snow and it kept me warm. https://weatherwool.com/pages/the-science-of-wool

You have some intriguing tech ideas. I was kind of inspired by a thread on the falcon project that got me thinking about AI/ML solutions for bigfoot drone surveillance. After spinning some cycles on it, I think it would be a lot of nerdy fun to try and implement. I have no plans to do it because of the time it would take to do properly. There is an enormous amount if dedicated tech that would need to happen. It seems easy to rig up thermal imagers on a drone and have it fly around and send quality images back to storage for analysis, but there is an expense of course and it would actually be difficult flying conditions over potentially difficult terrain. First challenge is flying a drone around at night without any BF targets hidden by obstacles just to understand the flight conditions and how to maneuver in difficult terrain. So many things can go wrong but if you can get that far, I was thinking a following phase would be making it an open source project and get community help to create the ML annotations. Finally, the hardest part is finding BF targets and trying to train your annotations to identify the proper targets on the ground, at night, in rough conditions for you and the drone. I see it as a lot of work over several years possibly to figure out the nuances.

My untrained and undereducated $0.02 speculation on the question of why is that there's too much land, too few of them and too few of us. All our ducks have to be in a row, we need to roll a nat 20 and they need to roll a nat 1, just to be in the same acre. They need to lose a saving throw just to be seen, let alone filmed. (Nerdy analogies) My theory is when random person sees one, it's that bigfoot's worst day, and that hiker's best. I think this because natural selection put us on divergent evolutionary paths, and we are far enough apart To be alien To each other. close enough to share diseases, but far enough apart that our common ailments are debilitating diseases to them, and one of them unlucky enough to catch our cold could kill their whole tribe. To improve our odds, I'm a technology guy, and lazy, I want force multipliers. Thermal is one, drones, drones with thermal, silent drones with thermal, drones shaped like birds, who fly programmed patterns at high altitude, and transmit images to a filtering site where humans review each frame, non-IR illuminators, non-IR triggers to cameras (maybe pressure plates or just a different frequency of radiation on the triggers to trail cameras, that are invisible to all mammal species), mechanical triggers like black thread, passive systems like remote cameras that are sound activated or long range lenses that are hundreds of yards away, "shot spotters" spaced out over an area passively listening for knocks or vocalizations that communicate and triangulate, then launch a drone equipped with thermal and visible light cams, of course all this really translates to $$$$. My opinion is these tools are already in use and regularly work for government forces, yes I'm one of those who think uncle Sam knows and is either protecting or exterminating, not sure which or why. It's a conclusion I've drawn from decades of observation of the way this and UFOs are treated by .gov, very similarly. Nothing more.