2 ms·
I know this isn't really the point of the article, but IMHO it's more interesting... can you really measure water speed just by looking at surface turbulence? W
by solardev 4y ago
I know this isn't really the point of the article, but IMHO it's more interesting... can you really measure water speed just by looking at surface turbulence? What do you do about fast moving but deep streams, or underwater current channels, etc.?
- kutx989 4y agoI was wondering the same thing. I know surfers look for rip tides using the surface. I was hoping the article would be about computer vision tools for that
- ghaff 4y agoSort of. To a degree. In a real world environment things are never perfectly uniform but it certainly can be easier to gauge flow if there are things disturbing that flow or visible objects in the flow. Lacking references something like a tidal current can certainly be faster than is obvious. (And people obviously can get caught in rip currents that may not be apparent.)
- mhh__ 4y agoUnder some conditions yes but presumably if you have a smooth flow at speed s and another at s*1.1 then you'd struggle.
- peterhalburt33 4y agoI was just about to say, I was slightly disappointed to find that this wasn’t an article about a novel technique to identify fast moving water (akin to the fact that we can identify the temperature water by the sound it makes when it is poured).
- solardev 4y agoIt's one of those interesting examples where Hackernews's dual identities conflict... it's half "interesting engineering stuff" and half "startup culture stuff". In this case, the article's business stuff made my eyes glaze over, lol, but that random thumbnail and metaphor they used was really interesting. I wish I could find the source of that image... unclear if it's just a mockup that NFX made for the blog, or if there's an actual machine vision water flow classification project somewhere. My Googling is really failing me this time. I can find articles (scholarly or not) on kayaking, water physics, machine vision... but not at the intersection of the above, and not any other instances of that image or lookalikes. Most of the AI projects out there are classifying aerial images from a bird's eye view, at ocean scale, not "how fast is this water that I'm looking at from shore" scale.