coming to you live

https://youtube.com/shorts/HX710NY-6z0?feature=share i did my best enjoy!😁❤️👍

tensorflow model maker is really neat! I'm just checking it out but it seems you can teach the models to recognize certain objects........ I like tensorflow even if it is the lite version. It's still awesome. 😁👍 hmmmm I wonder if Bill Shop guru ever looked at this I would be interesting to see a dronbot work shop video on it ......yep that sure would be cool .⛱️

@duce-robot

... it seems you can teach the models to recognize certain objects.

It seems that way but it doesn't really recognize for example a shoe as a shoe. All it does is give weights to features that have been found in thousands of images of shoes. That is not how we recognize shoes. We don't need thousands of images instead we are told that something you put on the feet to protect them from harm is called a shoe and maybe a few examples of such somethings to get the idea. The computer never gets the idea. That is why it might make some strange mistakes we would never make when there are unusual variations in the image.

This failure makes them unreliable for critical use. A slight modification to a stop sign may let a self driving car fail stop!

Under ideal conditions where images that are the same as the training set, the neural network might work fine. However the features it uses may fail catastrophically with a variation.

We never know what features are being used. There is the story of training a neural net to recognize military tanks by showing it pictures that contained a tank and pictures that did not. It seemed to work. However on new images it failed. Apparently the pictures with tanks in the training set were taken on a cloudy day and the others were not. It was recognizing a cloudy day, not military tanks.

@robotbuilder this was the tutorial I was talking about .

@duce-robot 

Being able to have them make models of your own objects is pretty neat although it would be rather tedious taking all those sample images.

It is giving a pattern a name or group name and you have to code what to do with that information.

Hi @robotbuilder,

RE your discussion starting, "It seems that way but it doesn't really recognize for example a shoe as a shoe."

  I think you are raising some valuable questions which deserves a thread or even an entire section in its own right, assuming there are enough participants with an interest .. which given the apparent interest in AI tools now coming on line there should be in the near future, if not already.

I would humbly suggest that a person recognising a "shoe" actually has at least two independent parts. The first is recognition of the 'intended use' of something that has characteristics such as hollow and 'foot-shaped and sized' , etc. suggesting it might be a something to wear on your foot. If your early years were spent in an 'urbanised' country, then the chances are your parents will have been fitting shoes on your feet from well before your earliest memories, so the connection will be 'ingrained' by learning from the 'start'.

You will also have been taught during those same early years that this object was called a 'shoe' .. if your home language was English .. but probably something different if your home language was not English ... so the connection between the name and the appearance of the object is separate from the connection between the usage and the appearance.

I would suggest that the contemporary AI systems which appear to be based on statistical matching maybe getting closer to the way that people, and presumably animals with brains, actually work than you suggest. Of course, I am only considering the similarity in terms of using a digital calculation machine being used to simulate a an immensely complex chemical processing 'machine'. Recently, the practical scale of digital calculation has begun to  approach the level of complexity needed to make a plausible simulator ... until now, the scale of calculation was just too small to do anything more than the simplest of 'brain' simulations.

Even assuming we can now make computational machines with sufficient complexity, so far we have very limited means of 'teaching' them ... getting them to 'read' endless petabytes of data swirling around on the Internet is much less direct and interactive than the infant being taught to put their shoes on ... so whilst the present machines may be far superior in finding research papers with keyword tags, their appreciation of the physical world (or universe) is limited by the focussed exposure.

I recall my first attempts at driving a car (with manual transmission) ... At the age of 17, I knew how it worked, and had done simple repairs to my father's car, watched my father drive for all of my life, but not actually driven it a centimetre. But none of that helped me coordinate steering, brakes, clutch, gears, etc. in the required manner .. that required a lot of 'hands-on' experience - 'exposure' to the real task.

So, I would suggest the limiations of 'exposure' also applies to humans .. the difference being the 'exposure' limits are different. As part of an urbanised society of citizens with a lifelong access to 'shoes', it is not surprising we have little difficulty recognising them. However, until a few hundred years ago, many people thought the Earth was flat ... allegedly some still do ... but for a person who lived inland a few hundred years ago, and only travelled a relatively short distance in their lifetime from their place of birth, the reasons for coming to the conclusion of the Earth being flat seem perfectly reasonable.

Similarly, with your tank recognition story, of which I heard a slightly different version about 40 years ago. I am unclear about the countries involved so I will call them 'A', who also instigated the study, and their potential foe, I will call 'B'. Namely, the military of 'A' wanted a machine that could distinguish 'A' tanks from 'B' tanks. Hence, they fed a series of photos tanks from 'A' and 'B' into the image recognition machine they had built. However, the photos for 'A' tanks were taken under 'realistic battlefield' conditions, which were dark and dull, grey days, whilst the only pictures of 'B' tanks were from displays or other 'publicity' material, all showing the tank in its 'splendour'. So yes, as in your version of the story, they had developed a machine that could distinguish lighting conditions, but knew nothing about tanks.

However, to me, that is similar to the people of a few hundred years ago, who maybe farmed on some 'flat' fields and had no reason to doubt the whole Earth was flat.

The semiconductor industry is only just beginning to make machines with sufficient calculating power, and they are still much larger and power hungry than the chemical processing systems found in biological systems. Perhaps the 'computers' wil evolve to be less like digital calculating simulators and more like our 'analogue' brains ... of course, this is part of contemporary research ..., but maybe the bigger challenge is connecting these machines with the 'real world', so that they can learn about it directly.

Best wishes all, Dave

@davee true a shoe or hat I've  noticed comes up as wardrobe.

I was trying to respond but my cave man hands fumbled it. Lol both good and interesting points.

Posted by: @duce-robot

↑

@davee true a shoe or hat I've  noticed comes up as wardrobe.

Perhaps once you train it to recognize a wardrobe that won't happen?

Are you going to get the hardware requried?

Hi @duce-robot and @robotbuilder,

Thanks for your comments... I hope you found something of interest.

re:

Posted by: @duce-robot

@davee true a shoe or hat I've  noticed comes up as wardrobe.

Perhaps once you train it to recognize a wardrobe that won't happen?

I suspect this also implies that it has trouble differentiating between a shoe and a hat, so even if you can train it to recognise a wardrobe, it could still be thinking you were wearing a shoe on your head ... so you now have two more things to train it to recognise ... and if for every thing you train it, you realise it is actually confusing two other things, you could be on an exponentially increasing curve of new training assignments.

Perhaps I am extrapolating to the point of absurdity, as maybe the list of different things to be trained has a practical limit, but I have a feeling we are just scratching the surface at present.

Best wishes and good health to you both, Dave

@davee you must  take individual pictures of each thing  and label them and use colab to make a custom file but I haven't tried yet

https://drive.google.com/file/d/1YC_3GdCVST2OZk96eLBJLt9m1lyKphL-/view?usp=drivesdk 😀

Hi @duce-robot,

  Just to let you know your Google drive link responds with a "You need access" response.

Wishing you very best wishes for 2024, Dave

@davee ok I'll fix it

The duce robot talks? Enjoy 😊