8
Heard a research guy say BCI is 10 years behind where the hype thinks it is
It was at some virtual meetup last night and this one data engineer from a lab in Austin said most consumer headsets still cant filter out muscle noise reliably. Made me realize my little win of getting a 70% accuracy on my motor imagery test at home might actually be closer to 40% if I accounted for artifacts. Has anyone else tried cross-checking their results against an EEG lab's standards?
2 comments
Log in to join the discussion
Log In2 Comments
wesley8310d ago
lol yeah I had a similar reality check when I tried to replicate that one study from 2019 about imagined speech decoding. Spent like 3 months getting my headset to work and my results looked amazing until a friend at a university ran them through their lab's processing pipeline. Turns out my filtering was basically garbage and I was just catching eye movement artifacts. The gap between home setups and proper gear is way bigger than people admit. Its kind of wild how many hobbyists are out here thinking they're hitting 80% accuracy when really they're just reading their own blinks.
2
the_ryan10d ago
So what specifically in their pipeline exposed the eye movement artifacts yours missed?
7