1,455 karma · joined February 14, 2021
Using the median or including standard deviation seems like the only way to approach an even slightly useful picture.
[0] https://www.businessinsider.com/apple-has-7000-fewer-people-...
Biggest thing I can recommend here is don’t conflate tools with process.
Think you shared that picture of police brutality anonymously? Think again.
It’s simply not possible to do depth estimation like this without priors. That’s one of the serious limitations of such systems - you have to train on every class of object you don’t want to hit.
Of course the vision system is supposed to compensate for this, and it performs poorly on objects it doesn’t see often, like emergency vehicles.
Discussion: https://internals.rust-lang.org/t/rationale-of-apache-dual-l...