Somehow when i add observations via website, after loading the metadata they all get automatically identified as spiders even though they obviously aren’t. Don’t think i ever tagged the photos. How does it came to be like this and how do i fix it?
It might be related to the new feature that was added that tries to auto tag species based on the file names. What was the image file name for one of these?
I had the same problem earlier this week. All my obs were automatically identified as Bay Laurel and sometimes a random tree frog, as the photo metadata was being uploaded.
Filenames are the usual DCIM filename e.g. OB247700.JPG
“Ob” is a listed common name for Araneae, it could be matching that:
We’re taking a look and will do what we can to minimize this.
I changed it so it now only looks for taxa in the language you view the site in, files names with
ob in them shouldn’t match Araneae b/c no one views the site in Tzotzil. Same goes for
Jun as mentioned in this post.
May is still a problem b/c it’s always going to look up scientific names, and apparently some extremely annoying taxonomist thought it would be cute to name a spider genus May. And even if they hadn’t made such an annoying choice, apparently a decent number of Britons throw semantic specificity to the wind and refer to Crataegus monogyna as “may.” I could just add a rule that says “ignore May,” or I could do something more sweeping and ignore all names that are three letters long or shorter. Thoughts?
Somewhere there is a thread about place names that double as genus names. I assume they could be an issue as well if people have location names in the file name?
EDIT - thinking about it, I guess this would be impacted by how or if scientific names are counted vs the language you are running in, so it may not be an issue
I could live with that if it seems like 3-letter words are doing more harm than good. So I’ll have to type in a few Poa names – no big deal
Is this still causing problems for anyone? If not, can we close this?
This has been happening to me for about 36 hrs,maybe 48. I thought I had accidentally typed something into the uploader because an obs self-ID’d itself as Linum. (Extremely unlikely, have not seen one in 15,000 obs.)
But the next obs did the same.
Then my Cordyline australis (very common here,and common among my obs) were all identifying themselves as a Cordyline I have never heard of. EDIT I never use the name cordyline in my file names - I use the common name ti kouka or tikouka or tik…
It is a huge problem because I upload dozens of obs at a time and often leave the taxon tillI have done all the obs, then select those obs of the same species and ID them, etc. When i have done, I scroll through the uploader to find any missing text in the taxon field. If the taxon fields are filled automatically i will have great difficulty checking each one.
i just saw, haven’t yet read all of, this thread, and haven’t uploaded anything for the last 8 hours or so, so I don’t know if the probelm has been fixed,sorry, but didn’t want the thread to close without replying to your question:)
EDIT have now read this thread. My files have all sorts of names, incuding my own location codes (letters) and names of plants, either common orLatin, including plants the observation doesn’t contain- eg “near the oak”
UPDATE IT is still happening. It seems I click on the Field while I decide what to enter, and while I am not watching it fills. Perhaps I need to modify my practice.
Perhaps it has always been this way but for some reason the AI response has become quicker?
Plus 1 to Diana’s suggestion below
Perhaps we need a personal setting - to choose whether or not we want to use this. Depending on how we name our photo files.
I would definitely like that.
Thank you for all the responses. My issue is fixed so far, though i agree with Diana’s suggestions; i don’t rename my photo files and i preferably don’t want this problem again in the future
Still happening to me on December 1. I wanted to load 18 pictures, but 8 were automatically called “Striped Bass.” None were of fish. They were birds and plants. And when I review suggestions for a single picture, all the suggestions were of fish species. I exited the page.
It’s not clear which if any of the suggestions (ignore 3 letter strings or shorter, only use the language the user is running in, other, none) were implemented, but I’m gonna guess your photos had ‘rock’ in the name, and that is entered as an English language name for Striped Bass. If so, that will have triggered it as it would not have been caught by any of the above.
Folks, I want to help, but I can’t unless you tell me what files names are resulting in what taxa. You can also tell me observation or photo URLs and I can look up the file names from that. “Everything is a striped bass” is not feedback I use to fix anything.
I don’t know for sure what you are asking, but I’m not very computer oriented. I loaded two pictures that were automatically identified as “striped bass.” Here is the URL for one of them:
And yes, it was taken at White Rock Lake, which is in my file name.
my file names never have taxon info but often have location info. Can we not have a code string to use to tell iNat what the name indicates?. for example ‘LOC’ followed by word(s) will indication location description , ‘TAXA’ followed by a word(s) will give taxon info and everything else is ignored. one could even have compound names such as LOC+words+TAXA+words which would populate location description and species name.
This would not help me, as having automatic population of the Species field by any means is a problem. I have uploaded several obs with faulty IDs since this change was made, even being alert to the issue and checking the entered ID with more care than usual. The problem is that since I nearly always enter the ID manually, once that field is filled out I disregard it as i move onto Fields, Tags, Description etc. Its hard enough getting those entered accurately when looking at 20-30 obs inthe uploader, so I would really appreciate regaining control over this aspect of iNat.