A client of their firm has 20,000 images. Some of those images are very suboptimal. My family member was assigned to deal with this by making a list of all the suboptimal images for the client. "Ain't nobody got time for that!" They also, apparently, have no budget.
-
-
Show this thread
-
So he is hitting the books figuring out Python scripting, and described an algorithm which is plausible from a CS perspective but perhaps implausible from a "I think those are actual real-world photos and, in that case, you're getting into deeper waters than you know" perspective
Show this thread -
So now he is wondering if he should skunkworks this with Amazon Turk or push forward with the Python image processing plan. Ironically, either of these will make him *discontinuously* more dangerous at his core job than he is today, for being assigned a month of scutwork.
Show this thread -
Sidenote: he said across their client population they might have a million images, and they're considering hiring multiple people full-time for a years long project to process them. I chortled a little. "It's *a million* images; that's no joke." "No you don't understand."
Show this thread -
"What?" "If you successfully have a proof of concept which works for 20k images, you are almost certainly less than one day of work on solving for a million images." "That's 50X larger though." "Neither your computer nor Amazon will care."
Show this thread -
"How hard is this problem that a real company is thinking of throwing multiple people for years at to solve?" It would be a little mean to give to an early career engineer as a job interview problem, but is materially smaller in scope than a hackathon project.
Show this thread
End of conversation
New conversation -
-
-
Err... So I'm just going to admit that I'm not really following the story honestly. Seems kinda past my year of experience. Could you please give a quick synopsis of what you were trying to say?
-
Software is surprisingly powerful, even in one's first real use of it, even for someone who would not in a million years consider themselves an engineer, even at sophisticated old-economy companies. Because it makes humans much, much better than they were without software.
- 1 more reply
New conversation -
-
-
what does “suboptimal” mean? what are the images? what i would do as a first cut is try this: pick 50 (or so) good images from the bunch and see which ones are most like them (by some metric) and see if that is a good discriminator or not. i am just spitballing btw.
-
You'll forgive me if I don't answer this in detail. "They fail to satisfy a property which a human six year old could trivially identify."
- 1 more reply
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.