Add idea


«    September 2019    »










DateDate: 3-09-2019, 05:47

  Personal non-property and property rights for the invention belong to the author and protected under current legislation. 
  Many mobile application developers are trying to find that theme to implement ideas to real work that is innovative and new. Only then, in most cases, there is the prospect of growth and promotion at the global level. Today's presentation is an example. The administration platform "Innovator" attempt to thoroughly present and explain the idea of this mobile application. 
  So, very often, in ordinary life, during sleep, in the process of the meetings and events we feel are moments that we think we are getting somewhere and somehow these moments of life were felt. Very often, we try to remember and reproduce in their thoughts or statements on these points in the hope to predict future developments. Or feel some coincidences that seem to be completely random; coincidence meet new people, which was long-desired; even sometimes, as they say, some people that way, and that, for example, being late for the bus that would be party to a car accident. Sometimes it works, sometimes not, but still, such moments often come to us. And based on this, very often we perceive it as a sign that we deserve it and start it all divided and sometimes to teach the basics correct the lives of others. 
  Innovative mobile app "SIGN" will give its users conduct their written blogs and to tell virtual users on all of these occasions and moments in life where , in their opinion, they saw some signs on which occurred any event. They will describe a little story about how it happened and how it ended. For example, the author has provided such a case from real life - a young man when dialing on the mobile phone wrong by one digit. The tube was lifted by the girl that the young man talked, agreed on a meeting, met and eventually they got married and live a happy life. This young man said it was a landmark situation, when he accidentally wrong by one digit when dialing a phone number. Or another example - a young couple bought the car on the automatic transmission. At the intersection, at a red traffic light, the car somehow dies. Although motorists know that on the automatic transmission the car should not stall ever. Now, when turned on the green light, the car the young couple was not able to start the movement for some time, as it was time for a new winding engine. But this time, in front of their car, a tree falls. If they began to move in time, this tree fell on their car, and the consequences would be unpredictable. All of these experiences can be called some kind of life signs, which should look at. Sometimes our attention can lead to such marks can lead to disastrous situations and consequences.
  The users of such momentous situations in their written blogs may give the opportunity to their readers to take from them the information which can will need in their future real life. Discuss the personal written blogs will be available to any registered user on this mobile app. In the form of a corporate identity, each such new written blog will begin with the words - OH, IT's a SIGN!
  This idea is open to investment, and full of redemption.

DateDate: 3-09-2019, 05:43
American and Japanese developers have taught the neural network to create a color 3D model of a person from one or more photographs. The peculiarity of the algorithm is that it recreates quite qualitatively even the view from the back, which is not visible in the original image, say the authors of the article that will be presented at the ICCV 2019 conference.
Reportedly, the solution, called PIFu (Pixel-aligned Implicit Function), consists of two consecutive convolutional neural networks, one of which analyzes the source image (or several source images), detects a human body on it, and using a technique called Walking Cubes ”, Recreates a 3D model from it, and the second gives color to the resulting model.
To learn the algorithm, the researchers used the RenderPeople dataset, which consists of high-quality 3D-models of people obtained using photogrammetric scanners.
See also: OpenAI updated the “dangerous” neural network that writes texts
In the video published by the authors, you can see that the algorithm recreates the whole model quite qualitatively, including from the back. In addition, the video demonstrates that even the presence of three frames instead of one significantly improves the quality of the final model. Finally, the authors showed that in many aspects the algorithm copes with the reconstruction of the 3D model better than similar algorithms of other developers.
According to the researchers, their development is another step towards creating a methodology that allows you to extract 3D scenes from ordinary videos. In further works, engineers intend to train the neural network to recreate 3D models of various objects from photographs.
Source: ITC