Google's new view of the world takes two pics to make 'DeepStereo' 3D StreetView means Google owns one of the world's larger photo albums, so it's natural for Google to want to create a realistic 3D rendering of the world. That's the aim of a new bit of boffinry from the Chocolate Factory called DeepStereo. As the group led by Googler John Flynn explain in this Arxiv paper, they found existing …
If the EU goes ahead with the proposal that the copyright of a building's photographic image is owned by the building's architect, it will be interesting to see how Google's library of such images will fare in any legal challenge or any other prior-established library of such images...
That means that all the buildings of interest (the historical ones) would probably be okay as it would only cover the new stuff that there's little interest in. But if there is modern stuff of interest (say The Shard) I could see Google making arrangements with the architect.
Then the architect can take his building out of my photograph. If you construct a building in a public space you should have no expectation to the ownership of a captured image.
How's this going to work with TV news crews? Are the journalists going to have to wait for the lawyers to get clearance from the architect before they can show footage of a protest that happened to be outside some unsympathetic organisation? Think protests outside embassies etc.
Aren't there enough problems in the world, that really need sorting out by our politicians and civil services, without wasting time giving rights to architects that they never expected to have when they started out?
If an architect has copyright on images of her buildings, does that mean I can sue the government for taking CCTV images of my face? And that the RAF has copyright on that bomb crater in the Falklands? And is it infringement if reflected glass on a building creates new images of an adjacent building (I'm pretty sure I've read about a sculpture in the US that's basically a big spherical mirror and it's copyrighted). And, say I designed a Shard-a-like, copyrighted my drawing, but never built it. When the real one was built, could I have requested it be taken down? Will Paris be asking royalties from Las Vegas, Blackpool and Tokyo?
I love absurd laws! I'm pretty sure that, if I laid all the unthinking legislators in the world in a line, my steamroller would need more than one tank of diesel to reach the end.
Machine learning imagines the missing pixels....
You mean just like organic learning has been doing for countless millennia?
The human eye is a bit crap....okay very crap....okay, by comparison to camera systems, mind-numbingly crap. Your eye picks up a small part of the image you see and only part of that in any serious detail. Your brain stitches all the seen bits together into one image and enhances detail in the bits that don't have it. What you see has way more to do with post processing than with optical imaging....
To be replaced with more subtle, but equally incorrect, artefacts.
Technically, no. The process will produce artifacts that have a higher probability of corresponding to the missing data, so probabilistically they're less incorrect.
They're more misleading, because, as you say, they're more subtle. In information-theoretic terms, they increase the uncertainty, because they will be wrong precisely when there's more information entropy missing from that part of the scene, and they'll do a worse job of indicating that information is missing. So this process actually weakens precision while leaving recall the same, in information-retrieval terminology.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
