mainframed

Google street view is not illegal. People just have the right to request that Google blurs their house. And because so many people requested it in Germany, Google said it's too much effort and disabled it completely. They still drive through Germany and collect images to improve their navigation services.

Also, I don't understand the discussion: Why is limiting free speech contradicting privacy efforts? Allowing people to prevent that their house being displayed on the internet is completely independent of stopping people writing racist slurs on the internet.

In Germany it's not about some extremist scalar ideology of freedom or control. You have basic rights and in daily life the rights of people collide.

In the case of Google Street view, there is the right of Google to choose their work (Art. 12 GG) vs. the right for privacy (derived from Art. 2 Abs. 1 GG) of people. Courts then have to follow a defined procedure to weigh those rights against each other. Apparently, it was ruled that the additional effort of blurring does not hinder Google's right to follow its occupation.

In the case of NetzDG (the law which enforces big social platform to delete criminal posts, such as racial slurs, which you probably refer to), you have the right of free speech of the poster (Art. 5 GG) and the right of human dignity of the guy who receives human dignity. Additionally, Art. 5 GG defines boundaries for free speech:

"These rights shall find their limits in the provisions of general laws, in provisions for the protection of young persons and in the right to personal honour. (Art. 5 Abs. 2 GG)."

I learned how the constitution works in detail in uni, and I think it's beautiful. To me, this is a recipe to have a healthy democratic society. And a democratic society has to be able to defend itself against anti-democratic, racists. Otherwise, it'll just collapse sooner or later.

What do you mean with that, and why does the error type/"randomness" matter?

If I can reduce the error rate by 90%, but the remaining 10% are "random" (whatever that means), is that worse than not reducing the error rate?

Ok, sure. I was just focusing on the 2D detection task, since that seem to be the focus of the linked product page.

You only have to label the 3D Bounding Box once. Then you can automatically generate 2D bounding boxes for every frame of the video. So instead of annotating every frame with a 2D box, you only annotate once with a 3D box.

Though, I wonder if the whole hassle of relying on RGB-D sensor of a phone, copying from your phone and using a yet another annotation tool, is worth it, when you can instead use some tracking bbox annotation tool, which interpolates many frames. With those, you can even annotate moving and distant objects, which I would argue is even better for generalization (since the background changes).

But I bet there are some use cases/users which can profit from it.

So the DB app works with MicroG?

Because, when I tried running without Google Services, this was one of the few apps, which I really needed and which did not work. Annoyingly only because they used it for loading the map, which I didn't need anyway.

Was excited about seeing an open alternative to Gitbook.

But yes, not loading without Javascript is a showstopper.