Will Mind-Reading by Photos Be Possible in Ten Years?
A lot of high-level photo apps and algorithms regularly appear on market. They can determine what is depicted on photos and identify whether a picture is funny or not. Why do we need all these innovations? Will they change our future life?
News covering great scientific and technical discoveries no longer excites us. Scientists now can bring back sight, cure HIV and make bionic organs. Engineers hone smart house technologies to perfection providing them with remote or automatic control and sustainable resource consumption, which allow to decrease adverse effects on environment and forget about household chores.
In contrast to these scientific breakthroughs new photo apps and algorithms seem to be a child’s play for those who are bored with simple photography and want something new. But it is wrong. These innovations are the future of photography techniques, advertisement, marketing and other fields.
We prepared a review of the latest achievements in photo technology and asked an expert to evaluate them and to predict their future.
Find People Looking Like You
Surprisingly, a Russian face recognition startup called N-techlab beat Google and Peking University in the algorithm contest The MegaFace Benchmark last year.
In February 2015 developers launched FindFace, the world’s first dating site that can find people looking like you or someone else. The app works on the base of the vk.com (Russian social network) accounts. The accuracy of results is 70%.
FindFace is getting more and more popular. It had 100 thousand users at the end of February 2016.
Determine Location where Picture was Taken
The latest Google service is neural network PlaNet that can determine where you shoot a picture. Now it depends on geotags. But in future PlaNet will operate on different photographic details: landscape, colors, architecture, certain animals and plants.
The developers used 90 million pictures to teach the neural network to determine locations.
Gender, Age and Emotions
How-Old.net is a Microsoft app that can tell the gender and the age of humans in a photo. The app almost always determines gender properly, but not the age.
Moreover, Microsoft developed an algorithm to recognize emotions. There are 8 types of emotions: anger, contempt, disgust, fear, happiness, neutral, sadness, surprise.
According to its users and developers, the results of emotion recognition are better than gender and age results, but not ideal either.
By the way, Microsoft neural networks are 5 times more effective than neural networks by Google, Intel, Qualcomm and Tencent.
Determine a Dog Breed
As a part of the experimental projects, Microsoft introduced What-Dog.net, a tool to identify a dog breed. You can upload your own images or find them in Bing.
The app recognizes only popular breeds, and doesn’t know rare ones. And don’t try to upload cats, it will identify them as dogs anyway.
According to Mitch Goldberg, Director of Development of the Microsoft Research, they wanted to generate an algorithm that would work on a certain criterion, for example, a dog breed.
Due to researches by Virginia Tech, Toyota Technological Institute and Microsoft Research department, neural networks were taught to identify humor in photos.
Scientists asked one group of volunteers to create funny collages using clipart. Another group of volunteers evaluated them from 1 to 5 points.
Researchers used results to teach neural network to determine whether the picture is funny or not. If the program decides that the picture is not funny enough, it will add some elements to make the picture more amusing.
Face Swap Live, a new iOS app that allows to switch the faces of peo
— Why do the world famous companies spend time on creating neural nple in a photo or video in real time. The app swaps faces of two people or combines one face with any image from the Internet.
Estimate Memorability Level
Scientists have learnt how many pictures a person can memorize for 4.5 min (four and a half minutes). They taught neural network and developed LaMem, a tool that estimates memorability index.
To test the tool, you have to upload a file to the website, as the result you will get a memorability map with the most notable marks of the picture, and its memorability index.
Scientists say that the tool memorizes photos as accurate as a human mind.
Why Do We Need It and What Will Happen in Future? Expert Opinion
Vladimir Gorshenin, Head of PressFoto Research Group
— Why do the world famous companies spend time on creating neural networks?
These algorithms are high potential. The global aim is to teach the machine to think and take the world as the human, to understand how people think.
All algorithms can be used in advertisement, marketing and photography. People create tons of content every day but they can’t process it quickly. The best solution is to give this work to robots. Moreover, inbuilt algorithms can identify if people like an ad or not, according to their facial expressions.
Also, such services can be built-in in smartphones to take pictures automatically, for instance, whenever people smile.
— Is it popular in Russia?
I think Yandex did some research. But in general, such activities are not supported financially. So only a few fanatics develop such projects. If you want to succeed in it, you need a lot of resources and funds.
— What can we expect in ten years?
Even now we can learn someone’s preferences and aptitudes by their fingerprints and find out chronic diseases by a palm print. I guess we’ll be able to read a person’s mind by his or her face soon!
Sure, if you are afraid of global surveillance, this technical news won’t make you happy. But you have nothing to worry about! Only if you are not a superstar and millions of fans don’t want to know details of your private life, or you are not going to rob a bank.
It’s time for Everypixel Studio
Published by admin, 26/01/2018
Russia Joined International Campaign for Intellectual Property Protection
Published by Natalia Zheleznaya, 05/05/2016
Small Size Images are No Longer Popular on the Web
Published by Asel Bakchakova, 05/05/2016
“Made by Author” event took place in Chelyabinsk, Russia
Published by admin, 25/04/2016
Will Mind-Reading by Photos Be Possible in Ten Years?
Published by Asel Bakchakova, 18/03/2016
Best PressFoto Photos of the Week
Published by Eugenia Ismagilova, 29/02/2016