this post was submitted on 23 Jul 2024
417 points (97.3% liked)

Science Memes

10304 readers
2516 users here now

Welcome to c/science_memes @ Mander.xyz!

A place for majestic STEMLORD peacocking, as well as memes about the realities of working in a lab.

Rules

  1. Don't throw mud. Behave like an intellectual and remember the human.
  2. Keep it rooted (on topic).
  3. No spam.
  4. Infographics welcome, get schooled.

Research Committee

Other Mander Communities

Science and Research

Biology and Life Sciences

Physical Sciences

Humanities and Social Sciences

Practical and Applied Sciences

Memes

Miscellaneous

founded 2 years ago
MODERATORS
[email protected]
[email protected]
[email protected]
417
Machine Learning (mander.xyz)
submitted 1 month ago by [email protected] to c/[email protected]
24 comments fedilink hide all child comments
 
you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 20 points 1 month ago (1 children)

I mean its just a matter of total available data points. The more images people take and upload, the more material they have to train their models. And obviously there will be way less people running around the tropics taking pictures.

[–] [email protected] 28 points 1 month ago (1 children)

It’s honestly way more about plant diversity. There are a million different plants in like a ten square mile area that all look exactly like an aloe and are related. The only way to differentiate them is by hyper obscure differences like their root structure and what their sap consists of.

You don’t even need to be in the proper tropics. Walk around San Diego with a plant id app and watch it spit out a different name for the same palm tree over and over because there are actually hundreds of varietals of palm with similar extremely complex identification processes. Some with toxic fruit and some with edible fruit that look the same.

[–] [email protected] 11 points 1 month ago (2 children)

I mean for those plants the model should be trained to spit out the next highest common denominator / family instead of the specific species. I would love to get a reply like "this could be any of the following species" instead of "im 23.232% sure that its this species"

[–] [email protected] 6 points 1 month ago (1 children)

Seek does that. That's the reason it's my go to ID app

[–] [email protected] 2 points 1 month ago

the full inaturalist app also has you upload the observation and likely get suggested IDs from other people

[–] [email protected] 3 points 1 month ago

I mean for those plants the model should be trained to spit out the next highest common denominator / family instead of the specific species.

Most people are going to take photos of the leaves, stem or at best the outside of the flowers. These are rarely conserved within families. You'll need the arrangement of the four floral whorls to name a family and expect any degree of accuracy. And that's assuming your plant is an angiosperm.