in

Chart groups 25,000 photos from 6.2 million images of Paris to show how similar we all are 


You might think you’re being creative with your Instagram snap, but chances are someone has take the exact same photo before you.

Now, one artist has used AI to visualise just how similar the photos we post on social media really are to those of our peers.

Moritz Stefaner’s collection of 25,000 photos reveal in incredible detail the crossover between individual shots of everything from tattoos to yoga poses.

The resulting clusters of pictures, taken from 6.2 million images of Paris geo-tagged on the social media site in 2017, are now part of a new exhibition in the French capital. 

Just how similar the photos we capture on Instagram are to those of others has been revealed in a new data visualisation created with the help of artificial intelligence. This image shows the entire 25,000 snaps analysed by aesthetics expert Moritz Stefaner who grouped them based on similarity using AI

Just how similar the photos we capture on Instagram are to those of others has been revealed in a new data visualisation created with the help of artificial intelligence. This image shows the entire 25,000 snaps analysed by aesthetics expert Moritz Stefaner who grouped them based on similarity using AI

Mr Stefaner is a ferman information aesthetics expert who has a background in cognitive science and interface design.

He used AI to analyse photos of Paris and organise them by appearance and image contents, revealing a web of interlinked niches and ‘microgenres’.

That includes similar images centred around a specific location, whether graffiti daubed on walls around the French capital to shots of its famous landmarks and even popular coffee shops. 

These niches cover scores of virtually identical shots of the Eiffel Tower, The Big Wheel Ferris – a Ferris wheel at Place de la Concorde – and a famous painting by Monet inside the Orangerie museum.

The image also span those focused on people, including shots of their hair and tattoos, that are surprisingly similar despite their idiosyncrasies. 

Clusters of pictures, taken from 6.2 million images of Paris geo-tagged on the social media site in 2017, show the connections between style and content - with thousands of near identical images discovered. This image shows the grouping of clusters of similar images

Clusters of pictures, taken from 6.2 million images of Paris geo-tagged on the social media site in 2017, show the connections between style and content – with thousands of near identical images discovered. This image shows the grouping of clusters of similar images

The map arrangement of the visualisation was calculated using an algorithm that is designed to create an ideal 2D layout so that all of the similar images are placed close together.

This highlighted just how close in appearance a large number of the shots within each cluster were.

That closeness is reinforced when viewing the individual images next to one another, with slight variations in each blending into the neighbouring shot.

One cluster centres around yoga, dancing and other poses poses performed by individuals, highly individual activities that still appear to melt into one another when viewed on aggregate. 

Another shows a whole section on macaroons and chocolates juxtaposed with images of human remains from the city’s famous catacombs.

In a written statement, Mr Stefaner said: ‘Today, we collectively and continuously document our city experience on social media platforms, shaping a virtual city image. 

While we may like to think that our carefully composed images reflect our unique personal tastes, the collection of 25,000 snaps charts in incredible detail just how much crossover there can be between individual shots. Shown here are shots of the Eiffel Tower

While we may like to think that our carefully composed images reflect our unique personal tastes, the collection of 25,000 snaps charts in incredible detail just how much crossover there can be between individual shots. Shown here are shots of the Eiffel Tower

The visualisation is the creation of German information aesthetics expert Moritz Stefaner, who has a background in cognitive science and interface design. Pictured here are very similar Instagram shots of the Big Wheel Ferris - a Ferris wheel at Place de la Concord

The visualisation is the creation of German information aesthetics expert Moritz Stefaner, who has a background in cognitive science and interface design. Pictured here are very similar Instagram shots of the Big Wheel Ferris – a Ferris wheel at Place de la Concord

The photos of Paris were analysed using neural networks organising them by similarity and image contents to reveal a web of interlinked niches and 'microgenres'. Many people posed in a similar way in front of this graffiti wall

The photos of Paris were analysed using neural networks organising them by similarity and image contents to reveal a web of interlinked niches and ‘microgenres’. Many people posed in a similar way in front of this graffiti wall

The visualisation includes similar images centred around a specific location, whether graffiti daubed on walls around the French capital to shots of its famous landmarks and even popular coffee shops

The visualisation includes similar images centred around a specific location, whether graffiti daubed on walls around the French capital to shots of its famous landmarks and even popular coffee shops

HOW DOES ARTIFICIAL INTELLIGENCE LEARN?

AI systems rely on artificial neural networks (ANNs), which try to simulate the way the brain works in order to learn.

ANNs can be trained to recognise patterns in information – including speech, text data, or visual images – and are the basis for a large number of the developments in AI over recent years.

Conventional AI uses input to ‘teach’ an algorithm about a particular subject by feeding it massive amounts of information.   

AI systems rely on artificial neural networks (ANNs), which try to simulate the way the brain works in order to learn. ANNs can be trained to recognise patterns in information - including speech, text data, or visual images

AI systems rely on artificial neural networks (ANNs), which try to simulate the way the brain works in order to learn. ANNs can be trained to recognise patterns in information – including speech, text data, or visual images

Practical applications include Google’s language translation services, Facebook’s facial recognition software and Snapchat’s image altering live filters.

The process of inputting this data can be extremely time consuming, and is limited to one type of knowledge. 

A new breed of ANNs called Adversarial Neural Networks pits the wits of two AI bots against each other, which allows them to learn from each other. 

This approach is designed to speed up the process of learning, as well as refining the output created by AI systems. 

‘Multiplicity reveals a novel view of this photographic landscape of attention and interests. 

‘How does Paris look as seen through the lens of thousands of photographers? What are the hotspots of attraction, what are the neglected corners? What are recurring poses and tropes? And how well do the published pictures reflect your personal view of the city?’   

‘To me, these very tight clusters of almost identical images became the most interesting aspect.

‘How often can people take the same photos? At the same time, each of them is slightly different indeed, and the continuous re-enactment of rituals and re-discovery of photo ideas has a comforting charm to it as well.’   

These images also span those focused on people, from poses performed by individuals to shots of their hair and tattoos, that are surprisingly similar despite their idiosyncrasies

These images also span those focused on people, from poses performed by individuals to shots of their hair and tattoos, that are surprisingly similar despite their idiosyncrasies

The map arrangement of the visualisation was calculated using an algorithm that is designed to create an ideal 2D layout so that all of the similar images are placed close together. Shots of newly-styled hair looking strikingly similar in style

The map arrangement of the visualisation was calculated using an algorithm that is designed to create an ideal 2D layout so that all of the similar images are placed close together. Shots of newly-styled hair looking strikingly similar in style

In a written statement, Mr Stefaner said: 'Today, we collectively and continuously document our city experience on social media platforms, shaping a virtual city image. 'Multiplicity reveals a novel view of this photographic landscape of attention and interests'. These images show similar yoga poses and stretches

In a written statement, Mr Stefaner said: ‘Today, we collectively and continuously document our city experience on social media platforms, shaping a virtual city image. ‘Multiplicity reveals a novel view of this photographic landscape of attention and interests’. These images show similar yoga poses and stretches

Mr Stefaner added: 'To me, these very tight clusters of almost identical images became the most interesting aspect. 'How often can people take the same photos? At the same time, each of them is slightly different indeed, and the continuous re-enactment of rituals and re-discovery of photo ideas has a comforting charm to it as well'

Mr Stefaner added: ‘To me, these very tight clusters of almost identical images became the most interesting aspect. ‘How often can people take the same photos? At the same time, each of them is slightly different indeed, and the continuous re-enactment of rituals and re-discovery of photo ideas has a comforting charm to it as well’

The visualisation and its clusters forms the central part of an exhibition running at the Fondation EDF in Paris from May to September.

Visitors will be able to navigate the map using a touch device as well as a physical joystick. Manual annotations help with identification of the main map areas – the larger themes linking clusters together.

‘Data was instrumental in arranging these contents in a human digestible way. How else would one scan and assemble hundreds of thousands of images into a coherent whole?’, Mr Stefaner added.

Projectors at the exhibit display images across three 1080p high definition squares arranged in a slightly angled tryptich structure. 

The visualisation and its constituent clusters forms the central part of an exhibition running at the Fondation EDF in Paris from May to September. Shown here is famous Monet painting inside the Orangerie ¿ where photgraphy is actually prohibited

The visualisation and its constituent clusters forms the central part of an exhibition running at the Fondation EDF in Paris from May to September. Shown here is famous Monet painting inside the Orangerie — where photgraphy is actually prohibited

Visitors will be able to navigate the map using a touch device as well as a physical joystick. Manual annotations help with identification of the main map areas - the larger themes linking clusters together. Shown here are pictures of coffee shops

Visitors will be able to navigate the map using a touch device as well as a physical joystick. Manual annotations help with identification of the main map areas – the larger themes linking clusters together. Shown here are pictures of coffee shops

This includes scores of virtually identical shots of the Eiffel Tower, The Big Wheel Ferris - a Ferris wheel at Place de la Concorde - and a famous painting by Monet inside the Orangerie museum, as well as street art and other graphics, including wall art (pictured)

This includes scores of virtually identical shots of the Eiffel Tower, The Big Wheel Ferris – a Ferris wheel at Place de la Concorde – and a famous painting by Monet inside the Orangerie museum, as well as street art and other graphics, including wall art (pictured)

The projected display zooms from a large overview map of the entire cloud of clusters, down to a gridded version of the cloud and on to each full grid of related images. 

This layering lets viewers explore the clustering structure across the entire chart as well as the contents of each. 

The approach used in its design makes sure that as little pixels change between the different transitions, leading to a more consistent zoom experience between each level of the graphic.

In summation of the project, Mr Stefaner added: ‘It has been my intention not to measure, but portray the city, but to portray it, using social media contents as material. 

‘Rather than statistics, the project presents a stimulating arrangement of qualitative contents, open for exploration and to interpretation — consciously curated and pre-arranged, but not pre-interpreted.’ 

Projectors at the exhibition display images across three 1080p high definition squares arranged in a slightly angled tryptich structure

Projectors at the exhibition display images across three 1080p high definition squares arranged in a slightly angled tryptich structure

The projected display zooms from a large overview map of the entire cloud of clusters, down to a gridded version of the cloud and on to each full grid of related images

The projected display zooms from a large overview map of the entire cloud of clusters, down to a gridded version of the cloud and on to each full grid of related images

This layering lets viewers explore the clustering structure across the entire chart as well as the contents of each

This layering lets viewers explore the clustering structure across the entire chart as well as the contents of each



Source link

The Rolling Stones announce pop-up shops across the UK

Blake Lively's Killer Met Gala Dress Took Over 600 Hours To Make