innovator

Add idea


Calendar

«    July 2019    »
MonTueWedThuFriSatSun
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
 

 

Advert

 

Payment

 

Advert

 

Authorization

Стартап

DateDate: 30-05-2019, 05:57

The AI predicts for producers from Hollywood what their box office money will have.
Artificial intelligence and machine learning neural networks are able to predict how popular and profitable a film that is still in production will be. New IT capabilities are already widely used in Hollywood, writes The Verge.
The film genre, script, cast and actors in specific countries are used as input. Then, using algorithms, the computer model calculates the box office, which the film can potentially receive.
 
If there are several variations of the same film - for example, with Emma Watson and Jennifer Lawrence in the main roles, then they can be compared and choose the best option.
“You can compare different versions of a picture individually or compare them in a package. For example, take both scenarios with Emma Watson and Jennifer Lawrence, and see which one for this particular film will bring the best results in different countries,” said the founder Cinelytic Tobias Kaysser.
Of course, there are also disadvantages - some predictions are too obvious, and then the use of expensive programs with AI is simply unnecessary. It is clear without them that a film with Leonardo DiCaprio will increase the picture's chances of great success at the box office.
A new neural network model was able to recreate a person's face by voice
In addition, the predictions are conservative in nature, because the model is trained on previously successful films. You cannot predict anything new or how a completely new actor will affect a film’s success.
But participants in the Hollywood market are confident that the algorithms cope much more accurately than experienced producers. The ScriptBook company notes that out of 50 analyzed films released in 2017-2018, the profitability or failure of the film was predicted correctly in 86% of cases.

DateDate: 29-05-2019, 05:58

Experts have taught technology to draw with the help of several million videos.
A group of American developers introduced the Speech2Face neural network, capable of drawing a portrait of a man according to his voice. Reported arXiv.org.
Models are able to draw an image, based on gender, race, age. The team engaged in the development of the Massachusetts Institute of Technology, which included Tahyon O.
Scientists decided to use AVSpeech for learning neural network. It contains about a million short videos of over a hundred thousand different people. In this case, the video and audio tracks are separated.
The accuracy of the technology makers determined according to three demographic indicators. Comparison was made of gender, approximate age and race of people from the original clips and the "conclusion" made by the neural network.
Were also identified the shortcomings of the model. She is not always able to determine age with an accuracy of ten years, and best of all depicts Caucasians and Asians. It is believed that the latter problem is related to the uneven presence of different races in the sample for training.
According to the researchers, their plans did not include a thorough copying of a person’s appearance. They sought to accurately identify gender, age, ethnic group.
In the photo: the original image, restored and "drawn" by voice.

DateDate: 28-05-2019, 05:58

Just a couple of weeks ago, the Japanese railway company East Japan Railway presented the Alfa X bullet train, which can reach speeds of up to 400 km / h.
As he wrote to ITC, talking about the newest passenger express, which will begin to regularly carry passengers only in the 2030 fiscal year, casually mentioning that the maglev trains (a special type of train held over the roadway, driven and controlled by the strength of the electromagnetic field) are very far away.
And so the other day, China Railway Rolling Stock Corporation (CRRC), the Chinese state railway company, completed the assembly of an ultra-high-speed maglev train capable of a speed of over 600 km / h.
Only “Hyperloop” vacuum trains can be faster, but so far they exist only on paper.
Regarding the latest development of the CRRC, the prototype of the superhigh-speed magnetic train was assembled at a plant in Qingdao, a port city in Shandong Province in eastern China.
With this maglev train, CRRC wants to connect Beijing to Shanghai, providing passengers with the fastest way to travel this route. The speed of modern high-speed trains on this route is limited to 350 km / h. The new train should overshadow even the flight connection.
Taking into account the time of check-in for the flight, the journey by plane will take about 4.5 hours, the trip by high-speed rail is about 5.5 hours, and the new high-speed maglev will take about 3.5 hours, ”the head of the Maglev development team noted in his statement -runks and Deputy Chief Engineer of CRRC Ding Sansan
In the near future, the CRRC will begin large-scale trials of the Maglev, after which the train will be launched into commercial production. This should happen as early as 2021.
It remains to add that the new Chinese prototype of the maglev-train will no longer be the first train, which has conquered the mark of 600 km / h.
In 2015, the maglev of the Japanese company Central Japan Railway developed a speed of 603 km / h on the test line in Yamanashi, thereby setting a new world record. Among the serial trains, the Shanghai Maglev is the fastest, having previously accelerated to 431 km / h.

DateDate: 27-05-2019, 05:56
Finnish design studio Aivan has created Korvaa headphones, in the production of which yeast and mushrooms were used.
This writes Dezeen.
In total, the construction uses six “components” grown from microbes.
Headphone earband is made of bioplastic, created from lactic acid, obtained from baking yeast.
The lining, in which the speakers are embedded, is made using the Trichoderma reesei mushroom - in industry they are used to produce pulp. From the fungus, a protein called hydrophobin is obtained, which is known to form a dense water-repellent foam. Aivan mixes cellulose with hydrophobin to obtain a stable, yet soft structure.
As a mat rial, which for additional convenience are covered with lining, a part of the mushroom called mycelium is used. It feels like skin to the touch.
Headphone speakers also cover the “case”, which is made from web-based protein, one of the most durable materials in the world. It is noted that this protein can also be used for the production of body armor.
The creators note that in its current form, the Korvaa is only a concept that cannot replace a familiar pair of headphones. His goal is to show the potential of technology called synthetic biology.


DateDate: 26-05-2019, 07:27
The robot was presented at the International Conference on Robotics and Automation (ICRA 2019) in Montreal (Canada).
Created by scientists at the Genoa Research Center (Istituto Italiano di Tecnologia, IIT), HyQReal is the successor to HyQ, a smaller model developed several years ago.
According to the publication TechCrunch, the robot has a length of 1.33 m and a height of 90 cm. Its weight is 130 kg, including a 15-pound battery that provides up to 2 hours of battery life. It is resistant to dust and water and can rise if it suddenly falls or falls over.
At the ends of the legs of the robot is a rubber "sole" that provides traction to the surface.
The video released by the developers showed that the HyQReal robot dragged the Piaggio P180 Avanti aircraft weighing 3 tons over a distance of more than 10 meters. The test took place at Genoa International Airport. The length of this aircraft is 14.4 meters, and its wingspan is 14 meters.
According to the developers, the robot can be useful for farmers and rescuers. Also, the device can be useful when conducting various inspections. At the moment, HyQReal is a research prototype.