Computer Vision Clean Up - Wiki (2.0)

General
1

I am making this forum topic because the old one (found here) was closed over a year ago and is now outdated (ish). There are new species that are being misidentified frequently, and the old ones in the archive are not being misidentified as much.

This forum post follows the same format and rules as the other one.

*** READ BEFORE REPLYING OR ADDING TO THIS WIKI ***

  1. This post is a wiki. Edit this post directly. Don’t reply to the topic with your species suggestions. The edit button is in the bottom right corner of this post.
    • If you are new to the forum you may not see the Edit button. Send one of the forum moderators a private message and one of them can edit your “trust level”.
  2. This is a “clean-up” list. Only include species that the community is having difficulty keeping up with, such as those with few people making identifications or generally overlooked groups. As a rough rule of thumb, the species should have at least ~75 observations that are currently misidentified.
    • It’s not for listing everything that’s commonly misidentified by CV.
    • It’s not for species being misidentified for reasons mostly other than computer vision.
    • It is for situations where the community is unable to keep up with the influx of new observations being misidentified.
    • It’s not for species that are already being well handled by the community reidentifying them as they come in.
  3. Use journal posts on iNaturalist, not this topic, to discuss specific taxa, identification tips, and find others to help reidentify these observations.
  4. This is not a topic to discuss the way computer vision works or how people use it. If you want to propose a feature request related to CV, ask a question about CV, or start a tangential discussion, be sure to first read the FAQ and search the forum for previous topics on this subject.

Common themes

  • Few or no closely related species included among the other computer vision suggestions
  • Difficult or impossible to identify to species, even by experts, based on typical iNat photos
  • Few identifiers with the expertise to add corrections at rank=species
  • Visually similar to many other species, but happened to accumulate enough observations to be included in the suggestions, while others did not
  • Broad distribution (multiple continents), making it difficult to “patrol the borders ”
  • Assigned (at some point) an overly broad common name that made it seem like the only organism known by that name:
    • “red velvet mite” for Trombidium holosericeum, “common green lacewing” for Chrysoperla carnea, “tarantula hawk wasps” for Pepsis
  • For plants: large amount of planted (casual grade) observations that remain incorrectly identified
  • Two or more species that are visually similar but have non-overlapping ranges, with each suffering misidentifications outside of range
  • Immensely prolific identifiers blindly accepting identifications far out of known range

Animalia

  • Annelida

    • Lumbricus: L. terrestris, etc

  • Arthropoda

    • Curculio glandium
    • Ecnolagria grandis
    • Heteronychus arator
    • Trombidium
    • Philodromidae
    • Rhynocoris iracundus, cuspidatus and rubricus

  • Chordata

    • Bos taurus

Fungi

  • Ascomycota

    • Cladonia chlorophaea
    • Podosphaera pannosa
    • Usnea: U. longissima, etc.

  • Basidiomycota

    • Agaricus campestris
    • Amanita vaginata
    • Chlorophyllum: C. olivieri, C. molybdites, C. rhacodes
    • Hygrocybe: H. coccinea, H. miniata, H. chlorophana, etc.
    • Hygrophorus roseobrunneus
    • Macrolepiota procera
    • Russula: R. sanguinea, R. cyanoxantha, R. emetica, etc.

Plantae

  • Angiospermae

  • Bryophyta

    • Ceratodon purpureus

  • Charophyta

    • Spirogyra

Rather than discuss things like tips for identification for each species here, create a journal post, and use the comments section. You can link to that after the entry here.

13 Likes
2

If there is any problem with me posting this topic, please let me know and I will take it down.

If there is a genus, order, family or any rank above species missing in the wiki, feel free to add it then the species by following the format (Assuming that something in it is being misidentified by CV)

  • genus or higher
    • species
1 Like
3

Red Russula mushrooms too. Many people put them to species (like R. emetica and R. rosacea) but they can’t actually be put to species without microscopy work.

2 Likes
4

I already added them.

2 Likes
6

It seems like a lot of fungi need cleaning up

1 Like
7

I have added Cladonia chlorophaea to the list. There are thousands of observations of it to species level, but at least in Europe it generally cannot be identified to species without chemical analysis (and iNat has a species complex to deal with the issue). Most of the records seem to come from North America though and while the checklist seems to say they have some of the cryptic species as well I wonder what is causing the discreptancy (and whether or not it actually can be identified from just a photo over there).

2 Likes
8

Yes, unfortunately many of them need a big clean up…

1 Like
9

I messaged jurga_li, and she said this : “Regretfully, there is no continent in which of all complex, only Cladonia chlorophaea s. str. would exist. Hence, everywhere it has to be identified as complex, unless picture of TLC plate with substances is provided along with the photo of lichen”.

1 Like
10

I added Spirogyra to this list. It is commonly found as this yellowish-green scum floating on ponds. However, it cannot be identified to genus level without microscopy, since there is also Mougeotia, Zygnema, and a few other genera.

2 Likes
11

So bump these observations up to family Zygnemataceae?

12

With the current taxonomy on iNaturalist, yes. However, recent research (e.g. Hess et al. (2022)) suggests that Spirogyra and Mougeotia are not that closely related. They put Spirogyra in its own order, which means the most specific taxon would then by Zygnematophyceae (aka Conjugatophyceae), unfortunately :(

1 Like
13

iNat must be a frustrating place for nature newbies who want to know everything’s name.

1 Like
14

Nature itself is a frustrating place for nature oldies (like me) who want to know everything’s name.

7 Likes
15

Thank you for this post. I have now moved my duckweed and redshank to higher classiffications pending expert review.

I believe there is a spelling error: Deschampsia ceaspitosa instead of ‘Deschampsia cespitosa’

1 Like
16

The spelling is “caespitosa”; “cespitosa” is a synonym.
https://powo.science.kew.org/taxon/urn:lsid:ipni.org:names:77240207-1

1 Like
17

CV consistently misidentifies Fragaria virginiana as Fragaria vesca. There are many hundreds, probably thousands of examples.

3 Likes
18

I suspect lichens are equally bad, maybe worse. The majority of IDs that I see are the suggestions from CV that the observer has accepted, usually based on a single posted photo. Few of these observations seem to ever be addressed and seconded or corrected by the lichen experts who participate here, I’d imagine for valid reasons.

2 Likes
19

I have reduced the number of R. emetica observations to 78;
image
(Unfortunately, the top observer has opted out of the community taxon)

im going to start on R. rosacea today
EDIT : That is a lot of observations in R. rosacea!
image

5 Likes
20

Thank you for that, I did wonder. Sadly the spelling with the a, which would tend to be my preference, brings up nothing when searching on iNat. Sorry for interrupting.

21

Just finished all the Eleusine indica.

Any other grasses which are particularly bad that also need work? Thinking about working on some Chloris next

3 Likes