What makes Paris look like Paris: an algorithm has the answer
Each city has a unique character, and many times we are able to identify them simply by seeing an image. In an increasingly globalized world, what makes a city like Paris recognizable at first sight?
That is precisely what many scientists have asked of Carnegie Mellon University (Pittsburgh) and the Ecole Normale Superieure (Paris). It seems that they have found the answer, but they have had to resort to an algorithm to decipher this enigma.
Large avenues, double-paneled windows, names of the streets in blue and green signs, street lamps and subway entrances with a characteristic design ... all these elements help Paris to have its own identity.
Thanks to this study carried out by both universities, we now know that in the French capital there are a series of elements that appear much more frequently than in other cities.
These objects that appear behind every corner, in every block, help Paris to be Paris. Analyzing the streets of the city through Street View, they began to collect elements that were repeated frequently. The following video shows how an algorithm carried out this whole process.
"If you see the Eiffel Tower you will immediately know that you are in Paris, the problem is that there is only one in the whole city"
Then they compared these repetitive elements with the data obtained in other cities (London, Prague, Milan, New York, Boston, Philadelphia, San Francisco, Barcelona, Sao Paulo, Mexico City and Tokyo), in order to clean up which components make it unique to Paris.
Emblematic points such as the Eiffel Tower or the Louvre Museum place us immediately, but the "problem" is that they only exist at one point of the capital.
Thanks to this algorithm we can identify what differentiating elements exist in most of the streets and its creators affirm that this technology can be used in other areas: "what aspects make Apple's products different from those of the competition?"
Lack of "uniqueness" in American cities
Interestingly, the algorithm had more problems when finding differentiating elements in the "American cities", resulting in different brands of cars or road tunnels.
In this way, they affirm that it is evident that the cities of the United States would have "a relative lack of singularity and stylistic coherence". They also emphasize that the car "reigns supreme" in its streets.
Image 11 12 18 18 05 Stuck
To demonstrate the value of their study, they asked different artists to make a sketch of a Parisian street (before and after showing them the results of the algorithm). As we can see in the image above. Although they were not very misguided, the second work is much more accurate.
That is precisely what many scientists have asked of Carnegie Mellon University (Pittsburgh) and the Ecole Normale Superieure (Paris). It seems that they have found the answer, but they have had to resort to an algorithm to decipher this enigma.
Large avenues, double-paneled windows, names of the streets in blue and green signs, street lamps and subway entrances with a characteristic design ... all these elements help Paris to have its own identity.
Thanks to this study carried out by both universities, we now know that in the French capital there are a series of elements that appear much more frequently than in other cities.
These objects that appear behind every corner, in every block, help Paris to be Paris. Analyzing the streets of the city through Street View, they began to collect elements that were repeated frequently. The following video shows how an algorithm carried out this whole process.
"If you see the Eiffel Tower you will immediately know that you are in Paris, the problem is that there is only one in the whole city"
Then they compared these repetitive elements with the data obtained in other cities (London, Prague, Milan, New York, Boston, Philadelphia, San Francisco, Barcelona, Sao Paulo, Mexico City and Tokyo), in order to clean up which components make it unique to Paris.
Emblematic points such as the Eiffel Tower or the Louvre Museum place us immediately, but the "problem" is that they only exist at one point of the capital.
Thanks to this algorithm we can identify what differentiating elements exist in most of the streets and its creators affirm that this technology can be used in other areas: "what aspects make Apple's products different from those of the competition?"
Lack of "uniqueness" in American cities
Interestingly, the algorithm had more problems when finding differentiating elements in the "American cities", resulting in different brands of cars or road tunnels.
In this way, they affirm that it is evident that the cities of the United States would have "a relative lack of singularity and stylistic coherence". They also emphasize that the car "reigns supreme" in its streets.
Image 11 12 18 18 05 Stuck
To demonstrate the value of their study, they asked different artists to make a sketch of a Parisian street (before and after showing them the results of the algorithm). As we can see in the image above. Although they were not very misguided, the second work is much more accurate.
Comments
Post a Comment