IA: Google analyzes and describes images automatically

[Christophe]

A further step forward at Google: the automatic image analyzer, and it works already not bad even if there are still bugs: http://www.01net.com/editorial/632782/r ... -l-humain/

Researchers at the American giant have managed to work together two neural networks so that a program can describe in one sentence and successfully the content of an image.

Comparative result:



Good or not in fine? One can ask the question ... at this rate in 50 years (what before?) Even the justice will be automated and robotic ... with the drifts that implies!

[sen-no-sen]

Good or not in fine? One can ask the question ... at this rate in 50 years (what before?) Even the justice will be automated and robotic ... with the drifts that implies!

In 50 years! Are you joking or what?


https://www.econologie.com/forums/demain-tous-chomeurs-t13279.html

The process has largely started, it may even be too late ...

[Christophe]

Yeah ... pfff

[Obamot]

Ay ay ay

Google's mission is " organize information globally and make it universally accessible and useful »

It's beautiful cynicism!

[Christophe]

Houla ... it's not with 0.5 M $ that they will go far ...

Google's mission is "to organize information globally and make it universally accessible and useful"

I thought that was Wikipedia's mission precisely?

[Obamot]

€ made the lawyer's salary at US $ 41'500.-

The fathers, they are horribly badly paid nowadays ...

[Christophe]

Uh I do not see how you get to this budget ...

What I wanted to say is that it remains a drop of water compared to what brews google no?

[elephant]

I do not know if it will win: it is piracy that is prohibited, not information on techniques and places.

Their thing is like wanting to forbid the sale of kitchen knives on the pretext that you can also kill with them.

[Capt_Maloche]

I found some old post notes for the forum
It's just to receive notifications on this topic that I heard about this weekend

In 20 years, the network will be so powerful that it can indeed take control, and since everything is managed from the net ...

When I think that it is a question of eliminating the learning of writing at school and going directly to the keyboard ... are sick

[Grelinette]

As usual, when we talk about innovations and technical progress, there is,

- on the one hand, the positive aspect of the imagination and skills of researchers who find and develop methods and algorithms to answer a need or solve a problem,

- on the other hand, all the dubious and negative uses that these new technologies allow.

The analysis and recognition of photos or drawings is a subject studied for a very long time:

Being able to use an algorithm to scan a database of images, or on the web, to find photos according to their content is very useful and interesting.

For example, look for all the clichés that represent a particular person (political, businessman, gangster, terrorist, ...), a place, a planet, an object, a vehicle, a specific material, etc., etc. ..

Moreover, search engines already do it very effectively for text. Why would images or photos be more problematic, especially since text is already linked to images (eg Facebook)?

Note also that recent news has shown, admittedly with some flaws, that one could quickly identify an individual from a video or a photo (see the French executioners who make war in Syria).

Would there be automatic facial analysis behind these latest cases? ...