I can strongly recommend the mobile app Flora Incognita for quickly identifying plants. It’s made by one of the German universities so it’s free, don’t have ads, and don’t collect your personal data. I even showed it to my mum and she loves it
I live next to a botanical garden which has climate-regulated glass domes with all sorts of non-native plants from around the world and it gets 90%+ of them right, but sometime you have to feed it different parts of the plant. PlantNet does look great too, I’ll have to check it out!
That's a nice anecdote, but I didn't just make up this fact. According to their paper [1], Flora has 4851 supported species, while PlantNet covered 30,261 at the time. Flora also tends to mix up cultivated/ornamental plants with wild counterparts, so your perceived accuracy may be deceptive if you're not a botanist. So yes, it's a cute app if you are a walking outside in central Europe, but it's nowhere near PlantNet.
Looks like they've both added a lot since then. If I select their worldwide species list in app, I get 38k for Flora and 85k for PlantNet.
I just gave them a try for some common trees around Costa Rica that I'm confident in identifying, and they both did very well, slightly better than PictureThis and iNaturalist, which I've been using for a while.
Besides the utility for users, Flora Incognita is also a massive crowdsourcing project for science: They use the GPS location of photos in combination with the identified plants to build a temporo-spatial database of confirmed plant sightings. (To my knowledge no link to any personal data of users.)
I use iNaturalist for plants and animals, mostly insects (and Merlin for birdsong recognition).
No ads, non-profit, contribute to research, etc. It also allows you to find out what others have spotted in your area. I like the range maps, the images across seasons, and other seasonal information.
The actual ML-based recognition is somewhat hidden, I think that part of the UI needs some work; it seems like it was made for researchers first then hacked for non-experts to use.
It is a pity that, while their focus is on openness and accessiblity, neither Flora Incognita, nor iNaturalist or PlantNet can be installed from the F-Droid appstore. I unfortunately don't use Google Play, so can't testdrive them.
Flora Incognita and PlantNet are not open source, so they cannot be published to F-Droid. However, iNaturalist is and the apk can be directly installed from the GitHub releases page.
PictureThis is an old favorite and still works quickly and quite accurately. I believe it’s a Chinese app though so do with that information what you will.
iammjm · · focus · HN ↗
dmos62 · · focus · HN ↗
sigmoid10 · · focus · HN ↗
iammjm · · focus · HN ↗
sigmoid10 · · focus · HN ↗
[1] <a href="https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.13611" rel="nofollow">https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041...
Eueudhsbsj32 · · focus · HN ↗
I just gave them a try for some common trees around Costa Rica that I'm confident in identifying, and they both did very well, slightly better than PictureThis and iNaturalist, which I've been using for a while.
darkwater · · focus · HN ↗
dmos62 · · focus · HN ↗
kamma4434 · · focus · HN ↗
max-ch · · focus · HN ↗
pbhjpbhj · · focus · HN ↗
No ads, non-profit, contribute to research, etc. It also allows you to find out what others have spotted in your area. I like the range maps, the images across seasons, and other seasonal information.
The actual ML-based recognition is somewhat hidden, I think that part of the UI needs some work; it seems like it was made for researchers first then hacked for non-experts to use.
<a href="https://www.inaturalist.org/pages/about" rel="nofollow">https://www.inaturalist.org/pages/about
stinos · · focus · HN ↗
rapnie · · focus · HN ↗
calmingsolitude · · focus · HN ↗
<a href="https://github.com/inaturalist/iNaturalistAndroid/releases" rel="nofollow">https://github.com/inaturalist/iNaturalistAndroid/releases
palata · · focus · HN ↗
otherayden · · focus · HN ↗
dyauspitr · · focus · HN ↗
jkmcf · · focus · HN ↗
Seek is 220MB...