4 ms·

Plantnet is cool, but there's a lot of missed potential for adding discrimination questions. For example, when the result isn't clear-cut, it could ask you the

Plantnet is cool, but there's a lot of missed potential for adding discrimination questions. For example, when the result isn't clear-cut, it could ask you the standard identification questions and narrow it down to a specific species instead of making you guess based on photos.

agilob 4y ago
That's correct, there's plenty of incorrectly identified plants, but the app says 89% accuracy for the first result and 11% for the second, so it's hard to argue. I've many times encountered plants that don't in live in the UK but the app say they do.
- causality0 4y ago
  The high accuracy results are not always, well, accurate. Several times I've had an 85+ percent result then looked closer and gone "ah shit, the leaf pattern is wrong, these are alternating instead of side-by-side" or similar.
- 7952 4y ago
  Inaturalist uses GPS and date to help with this. So it shouldn't show you stinging nettles in November.
riffraff 4y ago
when I was in school we had something called "dychotomic keys" which were big books to recognize plants with binary questions. Does it have a leaves? yes/no go to page XXX/YYY Is the leaf oblong? go to page WWW/ZZZ Is the border smooth? go to.. I wish there was a plant identification app using _that_ system, it would be much more entertaining and educational to use.
- TheaomBen 4y ago
  I've been meaning to take a crack at a software version of this for insects/birds/plants, do you know of any good ideally libre sources for the data?
  - morphle 4y ago
    I'd be willing to help you. The largest dychotomic key [1] is partially digitised but we could transform it into a proper dataset. Copyright no longer applies to part on the Flora. There are many other sources. I remember dychotomic key in hypercard [2]. Scanning 81 million science paper pdfs will yield most of the data in [1] and [2]. It would be possible to get a grant to make these transformations and add them to Plantnet and iNaturalist. [1] https://en.wikipedia.org/wiki/Flora_Europaea https://en.wikipedia.org/wiki/Flora_Europaea [2] http://www.etibioinformatics.nl/faq/ http://www.etibioinformatics.nl/faq/
    - TheaomBen 4y ago
      I am away from home atm but very interested in making this happen. I'll look into it a bit and get in touch by week's end! Thanks for the links.
      - morphle 4y ago
        It is also possible to add DNA material to field observations. A DNA scanner cost around $1000 and attaches to laptops and cellphones so they could be used in the field. I'm not sure if amateurs can handle the software yet, but that could be fixed. The most progress could come from my speciality: data mining scientific papers. This requires a few hundred terabyte of harddisks and a fast computer plus 2000 hours of programming. This would require a minimum of $20K or more in grants, crowdfunding or donations but it would yield an enormous boost of data. Another addition could be scanning all the herbaria and botanical gardens, both photo's and DNA samples. Crowdsourcing by thousands of citizen scientist would be the way to go.
  - stevesearer 4y ago
    The US Department of Fish and Wildlife has a really useful service called the Feather Atlas for identifying bird feathers by their color, size, and shape: https://www.fws.gov/lab/featheratlas/idtool.php https://www.fws.gov/lab/featheratlas/idtool.php
  - morphle 4y ago
- abibibo 4y ago
  For woody plants, VTree might fit what you're looking for. It's a dychotomic key optionally augmented by your location data. https://dendro.cnre.vt.edu/dendrology/vtree.htm https://dendro.cnre.vt.edu/dendrology/vtree.htm
  - riffraff 4y ago
    Wonderful, thank you for sharing!

agilob 4y ago

That's correct, there's plenty of incorrectly identified plants, but the app says 89% accuracy for the first result and 11% for the second, so it's hard to argue. I've many times encountered plants that don't in live in the UK but the app say they do.

causality0 4y ago
The high accuracy results are not always, well, accurate. Several times I've had an 85+ percent result then looked closer and gone "ah shit, the leaf pattern is wrong, these are alternating instead of side-by-side" or similar.
7952 4y ago
Inaturalist uses GPS and date to help with this. So it shouldn't show you stinging nettles in November.

riffraff 4y ago

when I was in school we had something called "dychotomic keys" which were big books to recognize plants with binary questions. Does it have a leaves? yes/no go to page XXX/YYY Is the leaf oblong? go to page WWW/ZZZ Is the border smooth? go to.. I wish there was a plant identification app using _that_ system, it would be much more entertaining and educational to use.

TheaomBen 4y ago
I've been meaning to take a crack at a software version of this for insects/birds/plants, do you know of any good ideally libre sources for the data?
- morphle 4y ago
  I'd be willing to help you. The largest dychotomic key [1] is partially digitised but we could transform it into a proper dataset. Copyright no longer applies to part on the Flora. There are many other sources. I remember dychotomic key in hypercard [2]. Scanning 81 million science paper pdfs will yield most of the data in [1] and [2]. It would be possible to get a grant to make these transformations and add them to Plantnet and iNaturalist. [1] https://en.wikipedia.org/wiki/Flora_Europaea https://en.wikipedia.org/wiki/Flora_Europaea [2] http://www.etibioinformatics.nl/faq/ http://www.etibioinformatics.nl/faq/
  - TheaomBen 4y ago
    I am away from home atm but very interested in making this happen. I'll look into it a bit and get in touch by week's end! Thanks for the links.
    - morphle 4y ago
      It is also possible to add DNA material to field observations. A DNA scanner cost around $1000 and attaches to laptops and cellphones so they could be used in the field. I'm not sure if amateurs can handle the software yet, but that could be fixed. The most progress could come from my speciality: data mining scientific papers. This requires a few hundred terabyte of harddisks and a fast computer plus 2000 hours of programming. This would require a minimum of $20K or more in grants, crowdfunding or donations but it would yield an enormous boost of data. Another addition could be scanning all the herbaria and botanical gardens, both photo's and DNA samples. Crowdsourcing by thousands of citizen scientist would be the way to go.
- stevesearer 4y ago
  The US Department of Fish and Wildlife has a really useful service called the Feather Atlas for identifying bird feathers by their color, size, and shape: https://www.fws.gov/lab/featheratlas/idtool.php https://www.fws.gov/lab/featheratlas/idtool.php
- morphle 4y ago
abibibo 4y ago
For woody plants, VTree might fit what you're looking for. It's a dychotomic key optionally augmented by your location data. https://dendro.cnre.vt.edu/dendrology/vtree.htm https://dendro.cnre.vt.edu/dendrology/vtree.htm
- riffraff 4y ago
  Wonderful, thank you for sharing!