The latest NN based approaches still fall short of ORB-SLAM - which is considered the state of the art (in terms implementations). However, it still has only 70% success rate in loop closures.
We tend to measure our accuracy with loop closures and slam graph consistency over time.
Can you talk more about what models of geography you're referring to? In school in the US, I was always taught there were 7 continents, including the North and South Americas as 2 separate ones. As far as I remember, there was never even a mention of other models apart from maybe some historical understanding before lands had been fully explored by Europeans.
This, like history, is one of those dogmatic aspects of state education where only one side get taught and implicitly assumed to be the only one. Geography books need one big asterisk in so many sections
Is there such a tool for use with python? preferably matplotlib compatible. I often find myself needing to plot some data to get a quick idea of what it looks like, and my work is generally on a remote host. Having to open a new ssh session with X forwarding is not difficult but a more frictionless approach would be nice.
> without some kind of direct rendering going on, I can't see how this would work on a terminal emulator.
Nah.
You just dump the pixels one by one to the screen using a for loop. You can even do some cpu math with each pixel if you want, and at the end you call XPutImage with the raster (or whatever your toolkit offers). This runs at more than 30fps in fullhd on a 10 year old cpu.
Notice that xterm is blazingly fast and runs entirely on the cpu. Think about it, the plots that this iterm2 demo shows were already done in realtime in the late nineties using the first pentium cpus.
The GPU is not really needed for the basic stuff a terminal emulator does. Of course, if you want to do crazy stuff like scaling by enormous factors, or ultra-fast scroll without skipping tricks, you'll need GPU support.
I guess there is no problem? You only need the gpu to compute fancy 3d graphics in high resolution. For normal terminal stuff and low resolution plots, anything should work.
If you use euporie [1], you can draw plots in a Jupyter notebook in the terminal using matotlib and friends, and have them displayed using terminal graphics.
