Difference between revisions of "DT's Kahn Presentation 4/21/16"

From dftwiki3
Jump to: navigation, search
(Reconstructing the World in Six Days)
(Reconstructing the World in Six Days)
Line 22: Line 22:
 
* '''Web site''': [http://cs.unc.edu/~jheinly/reconstructing_the_world.html http://cs.unc.edu/~jheinly/reconstructing_the_world.html]
 
* '''Web site''': [http://cs.unc.edu/~jheinly/reconstructing_the_world.html http://cs.unc.edu/~jheinly/reconstructing_the_world.html]
 
* '''Data Set''': The 100 million image dataset used in this paper was obtained from Yahoo: One Hundred Million Creative Commons Flickr Images For Research, Yahoo Flickr Creative Commons 100M  
 
* '''Data Set''': The 100 million image dataset used in this paper was obtained from Yahoo: One Hundred Million Creative Commons Flickr Images For Research, Yahoo Flickr Creative Commons 100M  
 +
<br />
 +
==Mike Tyka, The Art of Neural Networks==
 +
<br />
 +
(With thanks to MJ Wraga for discovering this video)<br />
 +
From the YouTube page:  "Mike Tyka, who is both artist and computer scientist, talks about the power of neural networks. These algorithms are capable to transform computers into artists that can generate breathtaking paintings, music and even poetry."
 +
<br />
 +
<center><videoflash>0qVOUD76JOg</videoflash></center>
 +
<br />
 +
* [[Image:weirdThingDeepDream.png|center|500px|link=http://i.imgur.com/5NlWLKg.gifv]]
 
<br />
 
<br />

Revision as of 12:42, 17 April 2016

--D. Thiebaut (talk) 13:37, 17 April 2016 (EDT)



Videos


Zittrain on Wikipedia



Zittrain.png


Reconstructing the World in Six Days


Researchers at the University of North Carolina have processed 100 million images from Yahoo and used the geolocation information of the camera, along with the image, to recreate a 3D model of various buildings. Once the 3D object is reconstructed, computer software can be used to "move" around the object. The video shows such reconstructions. 8 min 30 sec.


  • Reconstructing the World* in Six Days *(As Captured by the Yahoo 100M Image Dataset) (CVPR 2015),
  • Authors: Jared Heinly, Johannes L. Schönberger, Enrique Dunn, Jan-Michael Frahm
  • Presented at the Computer Vision and Pattern Recognition conference (CVPR), 2015
  • Web site: http://cs.unc.edu/~jheinly/reconstructing_the_world.html
  • Data Set: The 100 million image dataset used in this paper was obtained from Yahoo: One Hundred Million Creative Commons Flickr Images For Research, Yahoo Flickr Creative Commons 100M


Mike Tyka, The Art of Neural Networks


(With thanks to MJ Wraga for discovering this video)
From the YouTube page: "Mike Tyka, who is both artist and computer scientist, talks about the power of neural networks. These algorithms are capable to transform computers into artists that can generate breathtaking paintings, music and even poetry."


  • WeirdThingDeepDream.png