We planted some spinach last year, they grew, we harvested them and died since they were out of season. But as we speak I checked the garden and these grew in place of them, Bird's Nest Fern is what they're called apparently and I'm wondering if they're edible or not since they were planted from the spinach seeds we sown last year. Thanks in advance for the help!
Gemini thinks it's peony...
Note: Gemini is usually very good at identifying plants but some plants are notoriously difficult for anyone to pin down except horticulturalists (e.g. experts). For example, Pothos vs. Heartleaf Philodendron or more dangerous, Queen Anne's Lace vs Poison Hemlock. AI also has trouble identifying grasses (in general) because many can look the same from a picture. Even close up! Yet a human (in person) would be like, "these two grasses are nothing alike!"
It's all about pattern matching. If two leaves from two different plants have the same appearance and growth pattern, AI can easily get it wrong. In general, though, it usually does a really good job.
PictureThis! is the best AI plant ID I've found (not paying how much they want for it though). iNaturalist's AI is pretty good too (and real experts will also help ID stuff). I haven't tried Gemini. I tried ChatGPT and it sucked. Google Lense also sucks.
I have found PictureThis to be surprisingly accurate, even for just the tiniest shoots or a tree off in the distance. And it's perfectly usable for free once you close the subscription screen. (But I don't use it for anything I plan on eating.)
I have no idea what powers their identification. It must be some kind of machine learning because I can't imagine any other kind of image recognition data set being big and labeled enough.
Edit: it doesn't say on the app page but their social media accounts are all named PictureThisAI so I guess it is AI
Yeah, it's likely a more "traditional" image classification model (convnet or vision-transformer, not generative AI).
People are ridiculous for downvoting this comment. It is useful information on the topic. Ai output should never be believed by itself, but it can be a helpful tool as long a human verifies it
The anti-AI sentiment on Lemmy is quite extreme. To these people, all AI is bad. Even when it's useful. Even when it's saving lives. Even if AI was the world's only hope of solving global warming, they'd still unplug it without skipping a beat.
Because none of that matters to them. It's a religion. A beleif that's not founded in reason. They hate it. The very concept of it.
If all AI was powered via solar panels in a desert with no water cooling necessary, they'd still hate it.
If all AI was perfect at what it did and never made mistakes they'd still hate it.
If all AI was open source and free, running locally on everyone's PCs and phones, they'd still hate it.
I don't get it. There's literally no place for intelligent machines in their world.
Posting without verifying deserves downvotes. It's useless at best, and harmful at worst when it's wrong. Thus, downvotes.
Also, since it is verifiably wrong, extra downvotes.
So what about the person guessing?
Atleast they showed their homework. And even went so far as to provide information and context, saying it may not be right.
Fuck ai, but atleast be consistent instead of a hypocrite.
What about them? They're equally unhelpful.
Your Ultimate Gardening Guide.
Rules