Something scary: AI judged photo contest results

[sirch]

I've mentioned in other posts how AI can generate relatively sophisticated assessments of photographs, and while I don't "know" , I suspect you could fairly easily choose to train your AI bot on Ansel Adams photographs or Gary Winogrand photos etc and then use a Landscape AI bot trained on Ansel Adams pictures to assess a Landscape competition.

I suspect the difficulty is that AI might be artificial but it is in no way intelligent. As far as I am aware both Adams and Winogrand only produced monochrome work so an AI trained on their work would only judge B&W as "good". Also you need a large corpus of images with the full spectrum from good to bad and some way for the AI to learn what is good and what is bad.

The thing no one talks about in "AI" is the transfer function (or activation function) which is the mathematical function that calculates the output of a neuron from its input. These functions are programmed into the AI by a human and hence why AI in its current form is in no way intelligent.

 

[ancient_mariner]

These are being judged by an "AI judge" in real time with a rolling top twenty being displayed. First prize €500, and all you need to do is upload your files.

When I first looked, I thought they were nearly all awful, On a second look, the pictures have changed, and to my eyes, there are several that are OK, but the majority....

AestheticsCompetition

competition.excire.com

Excire who are behind this competition use AI to keyword photographs.

The blurb that goes with the competition says:

"The juror: our AI

You don't show your photographs to a large jury, but only to a single juror: Excire's artificial intelligence.

No personal preferences or tendencies limit its judgment.

Excire's artificial intelligence objectively evaluates your photographs.

To do this, it has learned from hundreds of thousands of photos what distinguishes good photos from bad ones.

This sample data was curated by numerous professional and ambitious amateur photographers and thus perfectly reflects aspects of image quality and aesthetics."

Is the site still functioning for you? I've tried on a couple of different browsers (one without my usual security measures) and also without the tracking data in your link, but there are no images being shown.

 

[Nod]

Have to say that I prefer photos that are on the rare side rather than the horrendously overcooked ones that appear to be A"I"'s choice at the moment!

 

[AndrewFlannigan]

These functions are programmed into the AI by a human and hence why AI in its current form is in no way intelligent.

Agreed.

I've had conversations with a psychologist who specialised in neural development. He argued that no machine system could be genuinely intelligent unless it was self booting, like humans. Perhaps it's not surprising that he had read Robert Heinlein's novel, "The Moon is a Harsh Mistress", which was written in the 1960s. One of the two major characters is a computer, which "becomes intelligent" rather than being programmed to appear intelligent.

My aquaintance said that he thought Heinlein, albeit writing long before any of the tools existed to to test the theory, had hit on the key difference between intelligence and an appearance of intelligence. In the psychologist's opinion, real intelligence depends on constant learning about both objects and their context plus the creation of self awareness, which he postulated as a multitude of feedback loops that were constantly increasing, as we experience stimuli.

I'm pretty sure that this is a rabbit hole, down which we could lose ourselves for hours!

 

[myotis]

I suspect the difficulty is that AI might be artificial but it is in no way intelligent. As far as I am aware both Adams and Winogrand only produced monochrome work so an AI trained on their work would only judge B&W as "good". Also you need a large corpus of images with the full spectrum from good to bad and some way for the AI to learn what is good and what is bad.

Well, that of course is the issue, AI isn't all that intelligent, which is part of the reason I'm opposed to using to judge photography competitions,

But, I was just suggesting that if you use it within certain limits it could be more useful than trying to make a model that tried to encompass all types of photographs.

Hence the suggested very limited scope ie one trained on Ansel Adams pictures assuming the judges were only interested in a winner that was an "Ansel Adams" type picture. And make use of some appropriate rule based decision making.

I posted a link earlier comparing two approaches to using AI as a Capture One assistant. One used a regular AI that searched the Internet for answers, and the other was expert trained. The difference between them was striking.

I'm still not keen on it's use for this at all, but it might be useful if you are sifting through a million competition entries or 5000 wedding photographs, where you want to weed out those that stand no chance of winning/being presented to the bride. There are already AI programs that do the latter.

As an aside Adams had at least one book in colour and did commercial work in colour. He also worked with Kodak on their first digital camera (which I assume was colour), something he was very excited about .

The thing no one talks about in "AI" is the transfer function (or activation function) which is the mathematical function that calculates the output of a neuron from its input. These functions are programmed into the AI by a human and hence why AI in its current form is in no way intelligent.

I have no idea about calculating the output of a neuron, but does anyone actually believe AI is "intelligent". The stuff, I've seen is pretty impressive, while at the same time seriously flawed, including some very stupid mistakes. But the speed it can do things, would seem to make up for the rather mixed results it produces, as long as it's subject to human scrutiny

I confess to not knowing much about AI, but as I mentioned earlier I've built some Bayesian Probability models, and the outputs from the model is controlled by the quality of the data used to build it, how we choose to interpret that data, and the rules we end up creating. So lots of human intervention.

 

[myotis]

Is the site still functioning for you? I've tried on a couple of different browsers (one without my usual security measures) and also without the tracking data in your link, but there are no images being shown.

I've been using Chrome, and it's still working for me.

I've now also pasted the link into Safari and it's working as well.

The pictures are taking about a minute to load, but as I said my connection is very slow.

I assume others have been successful.

I can't think of what to suggest

 

[ancient_mariner]

I've been using Chrome, and it's still working for me.

I've now also pasted the link into Safari and it's working as well.

The pictures are taking about a minute to load, but as I said my connection is very slow.

I assume others have been successful.

I can't think of what to suggest

I wouldn't worry. Waited over 2 mins, different browser again, nothing doing.

FWIW there used to be a site called (IIRC) Lightbox to which one could upload photos. The images would arrive at the front of a stream, and could be voted on. Images receiving votes/likes were kept near the front of the stream, with those that had fewer recent likes dropping back to eventual obscurity. Pictures were required to be relatively small, and the only way to get attention was to make an image striking, bright and contrasty with plenty of colour and a simple, strong subject. Anything subtle, complex or finely detailed would quickly disappear.

 

[Ed Sutton]

I wouldn't worry. Waited over 2 mins, different browser again, nothing doing.

FWIW there used to be a site called (IIRC) Lightbox to which one could upload photos. The images would arrive at the front of a stream, and could be voted on. Images receiving votes/likes were kept near the front of the stream, with those that had fewer recent likes dropping back to eventual obscurity. Pictures were required to be relatively small, and the only way to get attention was to make an image striking, bright and contrasty with plenty of colour and a simple, strong subject. Anything subtle, complex or finely detailed would quickly disappear.

It works for me in Firefox. Is it just the slide show that isn't loading or the 'gallery' lower down the page?

 

[myotis]

I wouldn't worry. Waited over 2 mins, different browser again, nothing doing.

FWIW there used to be a site called (IIRC) Lightbox to which one could upload photos. The images would arrive at the front of a stream, and could be voted on. Images receiving votes/likes were kept near the front of the stream, with those that had fewer recent likes dropping back to eventual obscurity. Pictures were required to be relatively small, and the only way to get attention was to make an image striking, bright and contrasty with plenty of colour and a simple, strong subject. Anything subtle, complex or finely detailed would quickly disappear.

You aren't missing much, but strange nonetheless.

As I mentioned earlier it seems it's human nature that if you are only giving a millisecond to look at a photograph you choose those with the strongest psychological triggers over those that demand more attention.

Although a totally different subject, it seems that Ultra Processed Foods have done the same thing, by removing the nutritious value (mainly to increase shelf life and reduce costs) and increased the intensity of the physiological triggers that makes us want to eat more and more of it..

 

[Pound Coin]

What an extraordinary collection of mostly “fake” images. I think they need to reeducate their AI.

Something unsettling about the name of the company niggles away, until I realised it starts similar to excrement.

 

[AndrewFlannigan]

it seems that Ultra Processed Foods have done the same thing, by removing the nutritious value (mainly to increase shelf life and reduce costs) and increased the intensity of the physiological triggers that makes us want to eat more and more of it..

Surely tinned and dried foods have provided better nutrition, to those with less money in urbanised countries and this is a very good thing?

It's been a long time since we could support the majority population of the advanced countries with fresh food. The rise of factory food production, has gone some considerable way to ameliorate food poverty in these countries. It could be argued, I think, that factory made, low cost food has been a major factor in the great increase in life expectancy over the last two centuries...

Life expectancy at birth

Espérance de vie à la naissance, Europe, pays développés

www.ined.fr

 

[jerry12953]

Punk didn't stop me liking the stuff I'd been listening to before, it gave me something else to listen to. Sure there was a load of crap in punk, but there was a load of crap before punk. Different crap!

Punk democratised 'pop' music. "It's easy, it's cheap, go and do it!"

The problem with punk was that it stopped the record companies promoting and the bands producing anything other than punk. The sex pistols sounded quite dangerous when they first arrived but rather quaint now.

 

[jerry12953]

Are there some parallels in the process where the big photo competitions are judged, I wonder? As in - the first round or two are judged by people with rather limited photographic knowledge, before the "winners" go on to be judged by the actual judges?

 

[myotis]

Surely tinned and dried foods have provided better nutrition, to those with less money in urbanised countries and this is a very good thing?

Neither of these things qualify as Ultra Processed (even if some tinned food can do). The basis of the argument against UPF is that it isn't bringing better nutrition, but that it's associated with range of serious health issues, linked to diet, including malnutrition.

At the extremes, there are several metadata studies that suggest UPFs are probably more addictive and harmful than alcohol or drugs.

It's a interesting topic, and off topic, but to give a simple and simplified example, UPFs use artificial emulsifiers instead of eggs to bring particular characteristics to food. This is much easier for a food manufacturer to deal with.Artificial emulsifiers are easier to "manufacture", transport, don't go off and have consistent qualities. But, don't bring any nutritional value to the food.

Another valuable thing about artificial emulsifiers is that you have more control over the texture of food, and texture is a key trigger in making us want to to keep on eating something. And there are strong links between UPFs, overeating and obesity.

These are very simplified examples just to give a feel for the issue, and as with everything, it's a complex issue. I'm still weighing up the evidence, but it doesn't seem something we can afford to just ignore.

I cut out UPFs 10 months ago, I'm now eating things that I used to avoid because of lifetime of dieting e.g. lots of nuts, various minor health ailments have cleared up, and my yo-yoing weight has stabilised at 5kg less than the lowest weight I have been in the last 30-40 (maybe more) years.

Things I've noticed are, for example, I've lost the constant craving for food that I used to have and by buying biscuits without UPF ingredients, I no longer have to fight off a desire to eat the whole packet., I feel I've had enough after one or two biscuits. To be fair these biscuits are also much more expensive, but they also last a lot longer.

And to bring this back to the photographic point I was making, the thing about UPF is that its engineered to trigger all the physiological levers that make you want to eat more, And they have hgh fat levels with very low nutritional levels.

Similar psychological triggers have been shown we are drawn to bright super saturated images, and I was suggesting that given people's millisecond attention span, they trigger a like, and people have moved on before realising how awful they are.

Studies have shown that given a choice, people prefer eating foods with UPF, because they have high levels of the physical characteristics we like in our mouth.

 

[myotis]

Are there some parallels in the process where the big photo competitions are judged, I wonder? As in - the first round or two are judged by people with rather limited photographic knowledge, before the "winners" go on to be judged by the actual judges?

I suspect that may will be the case, and in my earlier posts I'v suggested their may well be role for AI. It's already used by wedding photographers to make an initial cull.

But in the Excire example, they obviously aren't doing this as they are making a big thing about their "AI judge" being objective and unbiased, which clearly isn't possible.

 

[jerry12953]

I suspect that may will be the case, and in my earlier posts I'v suggested their may well be role for AI. It's already used by wedding photographers to make an initial cull.

But in the Excire example, they obviously aren't doing this as they are making a big thing about their "AI judge" being objective and unbiased, which clearly isn't possible.

Clearly.....

 

[AndrewFlannigan]

Studies have shown that given a choice, people prefer eating foods with UPF, because they have high levels of the physical characteristics we like in our mouth.

That suggests to me that they are a good thing, if allied with healthy but cheaper base foods.

I've become increasingly skeptical regarding claims about flaws in modern factory produced foods or indeed, the general cassandric tendencies of modern media. Of course, in the original myth, Cassandra's prophecies were always accurate but Apollo's curse meant that no one believed her. Our modern version is that no one can be at all sure which claims of doom are realistic and which are fantasies.

This in turn brings us back to the original discussion. If human judges had a series of parameters which, correctly applied, would define the goodness of an image, then an AI should be trainable to do the same. However, is there such a bunch of parameters that can be used consistently? I don't think there is.

 

[myotis]

That suggests to me that they are a good thing, if allied with healthy but cheaper base foods.

An interesting idea. The current problem is that it;s mainly the cheaper base foods that are dominated by UPF manipulation.

I've become increasingly skeptical regarding claims about flaws in modern factory produced foods or indeed, the general cassandric tendencies of modern media. Of course, in the original myth, Cassandra's prophecies were always accurate but Apollo's curse meant that no one believed her. Our modern version is that no one can be at all sure which claims of doom are realistic and which are fantasies.

I'm sceptical about everything. but still risk assess my decisions based on the available science.

This in turn brings us back to the original discussion. If human judges had a series of parameters which, correctly applied, would define the goodness of an image, then an AI should be trainable to do the same. However, is there such a bunch of parameters that can be used consistently? I don't think there is.

I didn't think this was the original discussion, I thought the original discussion was about how daft it was to use AI to judge a photographic competition. Which so far, unless I've missed something, everyone has agreed.

 

[AndrewFlannigan]

An interesting idea. The current problem is that it;s mainly the cheaper base foods that are dominated by UPF manipulation.

Which I consider a good thing, from the point of view of those who have to watch the pennies

I'm sceptical about everything. but still risk assess my decisions based on the available science.

As should we all. The problem though, is that there is now so much evidence, that we have to take our best stab at what is most likely to be the best evidence

I didn't think this was the original discussion, I thought the original discussion was about how daft it was to use AI to judge a photographic competition. Which so far, unless I've missed something, everyone has agreed.

I think you've misread what I wrote. My point is that if a human judge worked to such a set of rules then an AI could do the same. In the absence of such rules, then indeed the AI is silly.

 

[myotis]

Which I consider a good thing, from the point of view of those who have to watch the pennies

I think we just need to agree to disagree. I consider, encouraging people to eat food that has a low nutritional value and high levels of components considered harmful to human health, is a bad thing..

As should we all. The problem though, is that there is now so much evidence, that we have to take our best stab at what is most likely to be the best evidence

That is indeed part of the problem, but can be part of the risk assessment process

I think you've misread what I wrote. My point is that if a human judge worked to such a set of rules then an AI could do the same. In the absence of such rules, then indeed the AI is silly.

I don't think I misread you, I just don't think the thread has been discussing how human judges might judge a competition.

 

[AndrewFlannigan]

I think we just need to agree to disagree.

Always a sensible decision.

I consider, encouraging people to eat food that has a low nutritional value and high levels of components considered harmful to human health, is a bad thing..

I agree with you there but cheaper processed food doesn't necessarily lack nutritional value.

That is indeed part of the problem, but can be part of the risk assessment process

Indeed,

I don't think I misread you, I just don't think the thread has been discussing how human judges might judge a competition.

I may have injected that myself but it seems relevant to the discussion.

 

[myotis]

I agree with you there but cheaper processed food doesn't necessarily lack nutritional value.

But we aren't discussing "processed food", we are discussing Ultra Processed Foods.

Processed foods can be valuable sources of food e.g (as you suggested earlier), tinned food (non-UPF*) and dried fruit, both of which you incorrectly thought were UPF, Your misunderstanding was why I took the time to try and explain what UPF meant. (* my qualifier)

I may have injected that myself but it seems relevant to the discussion.

You certainly did, and I agree that as part of a bigger picture it's relevant, but, not necessarily part of the original discussion.

I think it lacked much of a response because it's something we have discussed multiple times in the past. It's a complex and involved topic and I doubt any of our views on it have changed,

 

[skullfunkerry]

Some interesting posts in this thread. The important question though: has anyone entered any of their photos into the competition?

 

[myotis]

Some interesting posts in this thread. The important question though: has anyone entered any of their photos into the competition?

It did cross my mind, briefly, to enter something as far away as possible from the current top twenty as I could muster (which could be any of my pictues), just to see what happened.

As I suggested in another post, what we don't know what has been entered and how that has affected the choices available to the "AI Judge" I think it's been advertised on their Facebook and instagram pages as well as in their newsletter.

 

[myotis]

To maybe wrap this up. Although I have a copy of Excire, I didn't realise it had this AI judging feature built in, which I assume this what the competition is based on.

Excire: AI Image Management Software for Mac & PC

Excire - viele intelligente Funktionen unterstützen Sie dabei, Ordnung zu halten und die Fotos zu finden, die Sie gerade suchen.

excire.com

To save clicking the link, it just says:

Image aesthetics

1. With the help of AI-powered image rating, each photo is given an aesthetic score between 0 and 100.
2. Based on the aesthetic value, the images in a collection (or all images) can be sorted in ascending or descending order by value.

The FAQ gives this explanation:

How does the AI aesthetics rating work?​

People often ask what criteria artificial intelligence (AI) uses to assess aesthetics and thus distinguish between "good" and "less good" photos.
We recently asked our experts for an explanation.
"Our AI has learned from a large number of photos what distinguishes good and bad photos. This data was evaluated by professional and ambitious amateur photographers, resulting in an average of around 200 ratings per photo.
As part of a training process, our AI then learned to imitate the photographers' ratings. The corresponding neural network has successively adapted its approx. 25 million parameters so that it can predict the ratings of the training data well.
The beauty of this is that you can now apply the AI to new unknown photos, as we do in Excire Foto 2024 and our photo competitions.
The neural network has therefore well encoded in its countless parameters what distinguishes good photos from bad ones.
It has learned this completely on its own based on examples."

So it looks as if they have asked "professional and amateur" photographers to score photographs, and then the AI software has correlated scores with common features/patterns that t has found in the photographs,

So it seems we are at the mercy of the "professional and amarteur photograhpers": that Excire chose to do the scoring, and as to whether the Excire IA chose to use features for its scoring that had any relevance to why a particular photograph received a particular score.

If I can be bothered, but don't hold your breath, now that I know, I have this "AI Photo Judge" readily available on my Mac, I might get it to score some of my pictures, I just have to decide if I want to risk getting high scores

 

[myotis]

If anyone is interested, here is a link to the email announcing th wiiners.

ð¸View the AMAZING winners of our landscape photo contest (judged by AI)

 

[ancient_mariner]

If anyone is interested, here is a link to the email announcing th wiiners.

Do you think the winning images were created by AI?

 

[myotis]

Do you think the winning images were created by AI?

No, my interpretation of the rules (posted earlier) made it clear AI generated images wouldn't be acceptable and they made the point of needing the original raws of the winning entries to prove authenticity.

Given the product they are selling, I don't think it would make sense for them to accept AI generated images and risk alienating their customers.

 

[Pound Coin]

Does anyone here use this software?

 

[myotis]

Does anyone here use this software?

Yes I do, sort of, in that I have a current license and I've played around with it.. I had the previous release and updated to this one to see how much it had improved.

So far, I haven't found any of the AI driven keywording programs to be much use, mainly because they added too many spurious key words. Excire suggests the smallest number of keywords but those that it does suggest are more likely to be relevant..

It also provides a useful and fast stand alone database for a photograph collection, if you don't have an alternative. I haven't tried it, but apparently it's very good at using AI face recognition to find all the pictures you have of a particular person, once you show a picture of the face you are looking for. It has a lot of clever things like this in terms of searching, but I've only used it for keywords.

It also does other smart type searches e.g find all of my photographs with two people standing by a Lake with mountains. Again I haven't tried this but the reviews I've seen suggest this works well.

I mentioned in an earlier post, that I hadn't realised it had an aesthetic score feature.

I've no idea why it's relevant to the software and why they have gone to the effort of adding it. I can't think of any occasion when it might be useful.

 

[Nod]

No, my interpretation of the rules (posted earlier) made it clear AI generated images wouldn't be acceptable and they made the point of needing the original raws of the winning entries to prove authenticity.

Given the product they are selling, I don't think it would make sense for them to accept AI generated images and risk alienating their customers.

Just as an exercise, could you try giving the software one of your images with the instruction to process it to win the competition?

 

[myotis]

Just as an exercise, could you try giving the software one of your images with the instruction to process it to win the competition?

I'm not sure I follow this, The program doesn't have any "processing" tools. It's an AI assisted database program.

 

[Pound Coin]

Yes I do, sort of, in that I have a current license and I've played around with it.. I had the previous release and updated to this one to see how much it had improved.

So far, I haven't found any of the AI driven keywording programs to be much use, mainly because they added too many spurious key words. Excire suggests the smallest number of keywords but those that it does suggest are more likely to be relevant..

It also provides a useful and fast stand alone database for a photograph collection, if you don't have an alternative. I haven't tried it, but apparently it's very good at using AI face recognition to find all the pictures you have of a particular person, once you show a picture of the face you are looking for. It has a lot of clever things like this in terms of searching, but I've only used it for keywords.

It also does other smart type searches e.g find all of my photographs with two people standing by a Lake with mountains. Again I haven't tried this but the reviews I've seen suggest this works well.

I mentioned in an earlier post, that I hadn't realised it had an aesthetic score feature.

I've no idea why it's relevant to the software and why they have gone to the effort of adding it. I can't think of any occasion when it might be useful.

Thanks, that's helpful.

I don't use Lightroom and my collection of images, over 30 years or so, is in multiple different libraries, folders, on different disks and so on.

I think I'll check if there is a free trial.

 

[myotis]

Thanks, that's helpful.

I don't use Lightroom and my collection of images, over 30 years or so, is in multiple different libraries, folders, on different disks and so on.

I think I'll check if there is a free trial.

If you are on a Mac Neofinder is an excellent DAM and worth looking at

NeoFinder - Digital Asset Management Software

Discover NoFinder, the Digital Asset Management tool for macOS and iOS. Try the amazing NeoFinder today.

cdfinder.de

(there is a Windows alternative Abemeda, but it seems to run behind Neofinder in features and updates)

It's a bit old fashioned but very powerful and very fast. Not as polished as Excire and without the AI tools, but solid old fashioned DAM, with great technical support.

It does actually offer AI keywording with a choice of AI engines (google. Apple and another I can't remember) but I found them less than good. One of them would identify birds down to species level , but it was always to the American species that looked the most like the UK species in my photographs, rather than the actual species.

Neofinder allows you to put a search window in your system toolbar to give access to instant and very fast searching. With my database of about 50,000 images I can type in "egret" and almost as fast as it take me lift my finger of the return key (pressed to initiate the search) I have a screen full of thumbnails.

 

[Nod]

I'm not sure I follow this, The program doesn't have any "processing" tools. It's an AI assisted database program.

Sorry, that's me getting the wrong end of the stick!