Ben201000

17 hours ago, GotAU? said:

As for accurately mapping where a detector coil has been, here’s a paper describing using a PDA and a camera to precisely map out the coil position and it creates a low resolution outline image of the target as the coil scans it. The primary use for the system was intended for UXO detecting, but it may be useful for relict detecting as well.  They even use the system for object recognition, but it would be useful for small targets though.  As for hardware, I’d think any new smartphone has the capability to do this, as the researchers who made the system were using a basic PDA and camera to do it.

Thanks for the paper! 

11 hours ago, BrokeInBendigo said:

You're most welcome. I've a keen interest in this, so please excuse if I go on a bit long about this stuff.

Thank you all of that post. There was so much in it. I'm new to everything to do with detecting so all of it was extremely helpful. What are your ideas on other methods of acquiring the signal? Is multi frequency analysis something that could be used?

20 hours ago, Chase Goldman said:

To do this accurately you need to move away from GPS as it can only be accurate at best to within 6 feet.  Fractions of an inch matter in coil coverage.  I'm thinking a portable local solution.  Perhaps consisting of laser tracking of a coil target monitored and recorded with precise positioning data coupled/integrated with visual information using a drone or drones hovering over the search area.  Think golf ball flight tracking and virtual first down marker lines across the video display of a football field.

I was thinking an approach to this could be to use a similar approach to photogrammetry or biometrics. Only using image and no location data. The phone/drone/AR glasses would map key points of each image and an AI detects the location of the coil against the image.

I'm doubtful if this could work right now, but I 100% think it will sometime in the future. 

In regards to you mentioning that induction is maxed out, are there any new alternatives? Maybe technologies that pick up so much that they're practically impossible to turn into a human interpretable audio signal, but could all be feed into some kind of neural network? Is multifrequency something in this direction? 

Hi Jasong,

Thank you for your post, it's got a lot of really good insight. 

What are your thoughts on the longer post I made about training an AI at each specific site? 

I also love Alpha Zero! It's so exciting the things that are being made 🙂 

18 hours ago, Mike_Hillis said:

You should buy a metal detector and join us.   

All I want to see/experience/hear is the object's density as compared to a standard, with an audio output that tells me how close the signal response is to the standard.

Hi Mike,

Thanks! I potentially will end up doing so. 

It sounds like detectors can already tell you the density compared to a set standard, or is this not the case?

20 hours ago, GB_Amateur said:

This paragraph seems to contradict itself.  "...Not a huge amount of training data is needed."  Then the next sentence:  "...over millions of iterations it starts to learn."  (emphasis mine)

 

Thanks for your reply 🙂 

I had worded the sentence badly. It goes through the set amount of data millions of times.

I'll check out the detector with the GPS that you mentioned in your last post.

19 minutes ago, Chase Goldman said:

The problem to be solved is accurately tracking the coil as it is swung through the site.  GPS does not cut it from a precision standpoint, but use of a stationary drone that is tracking and tracing the coil may be just the ticket.

Thank you for the insight. Initially I can't think of any solution to this but I really like your thinking. I'll have a ponder and see if anything comes to mind. I imagine the surveying industry will be the first to solve some precise gps that can get down to consumer cost. They have a lot of really interesting things with lidar and AI but I haven't seen anything around gps. I'm sure other members on this forum have a lot more knowledge than me around it.

Once the accurate gps is solved and smart glasses become common I'm sure combining them wouldn't be impossible.

19 hours ago, Clay Diggins said:

First off a pet peeve- AI since 2015? Nothing like changing the definition to achieve the goal. You certainly mean neural network Weak AI? Alan Turing is rolling over in his grave.

Real Artificial Intelligence doesn't exist - it's only hypothetical at this point. Real Artificial Intelligence wouldn't require humans feeding training data to a program. What the public is led to believe is that intelligent Strong AI exists then they are offered Weak AI as proof of concept. We were employing Weak AI back in the late 1970's at TI - nothing new but the name.

Hi Clay,

Thank you for such an in-depth reply. I completely agree, AI is a buzzword and I didn't mean to imply any true intelligence. By 2015 I mean the rise of deep neural networks and things like cat breed classification from images, and Dall E, GTP 3 etc. 

19 hours ago, Clay Diggins said:

To answer your question the "AI" you are referring to is used in several metal detectors today. You can start with automatic ground balance. Signal acquisition and processing is where most of the development is centered today. GiGo applies to metal detectors just like every other real world system and there is a huge amount of garbage in these signals.

Is automatic ground balance something that is already 'solved' or would this be a useful application? 

19 hours ago, Clay Diggins said:

The best we can do now as programmers now is sit back and let the engineers discover a working physical system to get the data that will allow true metallic discrimination.

By this do you mean a physical system as in some kind of laser or radar etc rather than an algorithm?

I've replied a longer post to another member. I'd love to hear what you think about it.

Thank you 🙂 

1 minute ago, Chase Goldman said:

No, that is really easily sorted and works well right now.  Tracking algorithms work well.  I am sure AI could be applied but really you have a solution looking for a problem when it comes to ground tracking, to be frank.

Hi Chase,

Thanks for the feedback. Do you see any useful scenarios in being able to train between two samples of anything while in the field?

20 hours ago, Redz said:

I see this as being particularly useful for problematic hot rocks, especially if you can grab your sample on the spot. Perhaps even as a smart ground balance.

Hi Redz,

Excuse me for the extremely dumb questions, but are hot rocks rocks that read similar to metal? Are there certain areas where there are lots which makes it harder to look for metals? 

Hi BrokeInBendigo,

Thank you for both such in depth posts. I'll do my best to try and summarize what I understand from them and then add/ask anything more. 

First it isn't very useful (potentially impossible) to detect between metals. It is far more useful to balance the ground and reduce EMI. Is this because EMI produces a lot of noise and leads to digging holes without any metal in? Or is it mostly distracting and mentally tiring over the hours? 

Secondly the ground conditions can vary enormously, essentially making a single model that knows everything impossible. Do the ground conditions vary hugely within one detecting session or is it more in different geographical areas? 

Something I hadn't considered was the scale of the area. I.e would a very experienced detector walk through a 5 kilometre square area and be able to narrow down a few 100 meter square areas that are worth detecting? If so, could AI be useful attached to a drone that scans a 2km area with various sensors attached, and then tries to narrow a few 100m areas to look in. As I write this I realise getting training data for this would be extremely hard. 

I really like the ideas around having a screen showing the raw waveforms. 

It seems from everyone's response that conditions can be so variable that the AI would have to be trained on each separate detection session. This presents a few challenges but could be possible. YouTube/Spotify etc all train customised AI models for each person. Right now it's possible to take about 30 photos of a bird vs not-bird on a hike, and train an AI in about two minutes on a standard laptop, with about 90% accuracy (not a hugely relevant example, but this was unthinkable 7 years ago). Of course no one wants to carry a laptop around, but there are AI specific training chips which are tiny. One on the iPhone as mentioned, but also more specialised ones like the Google Coral and Nvidia Jetson Nano. There is also the possibility of sending the waveforms over to servers to quickly train, but I imagine there isn't any 4G where everyone is detecting. 

Latest thought: There is a small AI training chip (no larger than a smartphone), with a screen (probably use smartphone app for it). The raw waves from the coil somehow connect into the chip. There are 3 buttons on the screen (label 1, label 2, train AI). When someone reaches the specific site they begin by choosing the labels. Maybe metal vs not-metal, but it could be anything depending what the detector thinks could work. As you dig holes you click which label it turned out to be. After say 30 holes you click 'train' and wait about 2 minutes. It then starts giving you a real-time prediction in a percentage of which label it thinks the coil is detecting. 
In no way does this replace the detector, you'd still use your usual method, there would just be something extra to test against. The approach of this method would be to have little AIs that are personalised to each persons local knowledge, and these AIs are each persons property unless they decide to share them. They may find that it works well distinguishing between two local features and they use it for that. Or that training after 10 samples is handy but they only trust the results about 40% of the time. Or that 500 samples is required, but that's okay because they regularly detect on similar ground. Or they upload and combine data with others in similar ground conditions. 

My last thought was that to collect data wouldn't it be possible to bury a few known items and scan over them in the specific soil? Or on a larger scale bury lots of samples in an area and scan over at different heights, add more soil above and scan again etc? 

To clarify how 30 samples could possibly work. We could use transfer learning on an AI trained on millions of images (called ImageNet), and then it is taught the new samples. Some people have had a lot of success turning waveforms into images and then training them with this AI. 

Thank you so much for the Dall E prompt booklet! I've got access to it so will definitely have a read through. 

Also are you in Bendigo, Victoria, Australia? I'm in Melbourne 🙂 

Welcome,

I was wondering if anyone if using any form of artificial intelligence in their gold detectors? 

I'm an artificial intelligence (AI) programmer. Last night I was watching Aduk gold on Netflix and suddenly wondered if anyone was using AI? A quick google didn't return much. 

I'm not sure how familiar everyone is with AI, but essentially since around 2015 it's blowing away peoples expectations year upon year. Simply put, traditional computer programs require humans to code in rules that then lead to a result. AI does that backward, it takes the results and creates its own rules to get to that. 

To do this with gold would require the creation of training data to feed into the AI. I don't know anything about gold detecting but I imagine you'd bury some pieces of gold and go over it with the detector, then save the waveforms (or equivalent) onto a computer. These become the gold samples. Then also bury things that most often give the most false positives compared to gold, and save those waveforms. 

The exciting thing compared to a few years ago is that not a huge amount of training data is needed. It's possible to take huge AIs trained by Google and then teach them the new gold samples far quicker with a lot higher accuracy. The process of training is basically the AI guessing 'gold vs not gold', and over millions of iterations it starts to learn. It then can be saved and used in the field to give a percentage estimation of how likely it thinks something is gold. 

I don't know anything about gold detecting but this is how I would see it used practically. Am I right in thinking that a fair amount of time is spent digging up false leads? If it's not and most of the time is spent surveying the area then the AI isn't very useful. But if there is a lot of time digging up false leads, then if the AI could save someone digging up 90% of the false leads would this create a lot of value? 

I'm really interested in any thoughts that any of you have regarding this. 

Cheers

Ben