33 ms·

This already exists: https://www.inaturalist.org https://www.inaturalist.org

Symbiote 6y ago
The machine vision on iNaturalist is great, but it's not necessarily adequate for distinguishing between insect species -- there can be details that are difficult to photograph, let alone identify from a casual user's photograph. And we still need training data (i.e. identified by the expert taxonomists) for future work.
- xipho 6y ago
  I have been at talks by their devs. They candidly admit it will never meet the needs (accuracy) to address questions at this level of specificity. Some things will be extremely successful, many (most) will not. Think of biodiversity as a curve, with a long tail. Will AI on poorly taken images specimens work for the bell? Probably. Will it actually get at the numbers at the tail? Almost certainly not. This is largely because 1) getting at the tail requires intimate knowledge of where to find that biodiversity (the vast majority of iNaturlaist pics are shockingly close to civilization, where diversity may not be) and 2) intimate knowledge, often of internal features or other non-imagable data, so that one can actually record data that fits in the tail.
- bliteben 6y ago
  inaturalist really needs the funding to support videos honestly.
throwanem 6y ago
iNat is good, but crowdsourced, and anything but a panacea; for example, I'm very confident in my identifications of the Auplopus mellipes and Auplopus architectus metallicus individuals I encountered and photographed last year, but those (plus one braconid, of which I'm admittedly uncertain myself) are the only observations I have on iNat that haven't had any species confirmations from other users. (That's fair; I needed considerable literature review to develop that level of confidence, and I doubt most iNat users share the extent of my interest in wasps.) Another option is BugGuide (https://bugguide.net https://bugguide.net), which I've actually found more useful - it did me a lot of good figuring out those auplopids. It doesn't provide any sort of dichotomous key (that I've been able to find), but there are a lot of observations available, often with very high-quality images, which helps a lot when specific identification depends on subtle features. In general, though, hymenopterans and especially solitary wasps are just difficult. There are many genera and even whole families that are hard to tell apart with much confidence. One of the first things you notice in the literature is the importance to specific identification of male genitalia, which are internal, practically microscopic, and only even potentially useful if you have the good fortune to happen on a very rare male of a solitary wasp species - and even then, sexual dimorphism can be such as to make a solid male identification useless for spotting a female unless you actually observe them to be in copula. (Female mutillids, aka "velvet ants", don't even have wings!) All of which is to say, I would expect the more obscure solitary bees to pose the same sort of difficulty.
- cmehdy 6y ago
  Do you know of any place where one could simply contribute? Bugguide seems to be North America only, but while I stayed in a remote area of the Indio Maiz reserve in Nicaragua I stumbled upon a spider that I haven't identified in years and can't quite find any place to simply share the photo[0]. [0] unknown spider on a banana leaf in eastern Nicaragua: https://i.imgur.com/iFMJxtl.png https://i.imgur.com/iFMJxtl.png
  - Symbiote 6y ago
    Isn't this the primary purpose of iNaturalist?
    - throwanem 6y ago
      Yes, iNat is worldwide and that's the first place I'd try. That said, I wouldn't necessarily expect a whole lot. I'm not even an amateur arachnologist, but my sense is that spiders are about as complicated as hymenopterans, if not more so; especially outside well-known families like the salticids, it seems like it's not rare to find difficult identifications.
  - pvaldes 6y ago
    > unknown spider on a banana leaf in eastern Nicaragua: https://i.imgur.com/iFMJxtl.png https://i.imgur.com/iFMJxtl.png Without having the eye arrangement is not easy, but taking in mind the habitat and shape, a member of Ctenidae would be my first candidate Compare for example with Ancylometes or Ctenus. If not, we could start weaving further from there. Zoropsidae, Lycosidae or Sparassidae (check Heteropoda venatoria for example) would be other acceptable family candidates. Those last couple of legs seem very characteristic.
    - pvaldes 6y ago
      Definitely Araneomorpha Lycosoidea, so there are ten families of interest or so. We don't have the eyes so we'll never know. Cuppiennius appears often in banana leaves, but doesn't have that strange angle in the last legs (Maybe is an abnormal specimen or an effect of the camera?).

The machine vision on iNaturalist is great, but it's not necessarily adequate for distinguishing between insect species -- there can be details that are difficult to photograph, let alone identify from a casual user's photograph. And we still need training data (i.e. identified by the expert taxonomists) for future work.

xipho 6y ago
I have been at talks by their devs. They candidly admit it will never meet the needs (accuracy) to address questions at this level of specificity. Some things will be extremely successful, many (most) will not. Think of biodiversity as a curve, with a long tail. Will AI on poorly taken images specimens work for the bell? Probably. Will it actually get at the numbers at the tail? Almost certainly not. This is largely because 1) getting at the tail requires intimate knowledge of where to find that biodiversity (the vast majority of iNaturlaist pics are shockingly close to civilization, where diversity may not be) and 2) intimate knowledge, often of internal features or other non-imagable data, so that one can actually record data that fits in the tail.
bliteben 6y ago
inaturalist really needs the funding to support videos honestly.

throwanem 6y ago

iNat is good, but crowdsourced, and anything but a panacea; for example, I'm very confident in my identifications of the Auplopus mellipes and Auplopus architectus metallicus individuals I encountered and photographed last year, but those (plus one braconid, of which I'm admittedly uncertain myself) are the only observations I have on iNat that haven't had any species confirmations from other users. (That's fair; I needed considerable literature review to develop that level of confidence, and I doubt most iNat users share the extent of my interest in wasps.) Another option is BugGuide (https://bugguide.net https://bugguide.net), which I've actually found more useful - it did me a lot of good figuring out those auplopids. It doesn't provide any sort of dichotomous key (that I've been able to find), but there are a lot of observations available, often with very high-quality images, which helps a lot when specific identification depends on subtle features. In general, though, hymenopterans and especially solitary wasps are just difficult. There are many genera and even whole families that are hard to tell apart with much confidence. One of the first things you notice in the literature is the importance to specific identification of male genitalia, which are internal, practically microscopic, and only even potentially useful if you have the good fortune to happen on a very rare male of a solitary wasp species - and even then, sexual dimorphism can be such as to make a solid male identification useless for spotting a female unless you actually observe them to be in copula. (Female mutillids, aka "velvet ants", don't even have wings!) All of which is to say, I would expect the more obscure solitary bees to pose the same sort of difficulty.

cmehdy 6y ago
Do you know of any place where one could simply contribute? Bugguide seems to be North America only, but while I stayed in a remote area of the Indio Maiz reserve in Nicaragua I stumbled upon a spider that I haven't identified in years and can't quite find any place to simply share the photo[0]. [0] unknown spider on a banana leaf in eastern Nicaragua: https://i.imgur.com/iFMJxtl.png https://i.imgur.com/iFMJxtl.png
- Symbiote 6y ago
  Isn't this the primary purpose of iNaturalist?
  - throwanem 6y ago
    Yes, iNat is worldwide and that's the first place I'd try. That said, I wouldn't necessarily expect a whole lot. I'm not even an amateur arachnologist, but my sense is that spiders are about as complicated as hymenopterans, if not more so; especially outside well-known families like the salticids, it seems like it's not rare to find difficult identifications.
- pvaldes 6y ago
  > unknown spider on a banana leaf in eastern Nicaragua: https://i.imgur.com/iFMJxtl.png https://i.imgur.com/iFMJxtl.png Without having the eye arrangement is not easy, but taking in mind the habitat and shape, a member of Ctenidae would be my first candidate Compare for example with Ancylometes or Ctenus. If not, we could start weaving further from there. Zoropsidae, Lycosidae or Sparassidae (check Heteropoda venatoria for example) would be other acceptable family candidates. Those last couple of legs seem very characteristic.
  - pvaldes 6y ago
    Definitely Araneomorpha Lycosoidea, so there are ten families of interest or so. We don't have the eyes so we'll never know. Cuppiennius appears often in banana leaves, but doesn't have that strange angle in the last legs (Maybe is an abnormal specimen or an effect of the camera?).