Data Quality Assessment

Shouldn’t a line be drawn when a picture is so low quality that, maybe you can squeeze out a fairly confident identification but it has no business being regarded as ‘Research Grade’? Only been thinking about this stemming from contributors posting audio observations of birds that aren’t trimmed, with loads of other sounds and duplicating for different species. I guess the latter is just a matter for discussing with them - but I know some people might chose to confirm the ID making it a very low quality observation.

1 Like

No, such a line should not be drawn. If the organism is identifiable, then that’s an accurate record of it at that time and place. Also, keep in mind that what’s unidentifiable for one IDer may be an easy ID for another, and photo judging in general is discouraged by iNat.

As for duplicating for different species, that’s a great thing! It shows a more complete picture of organisms and their environment.

10 Likes

It doesn’t matter how many species you can hear, one sound from target species is enough, take in mind you also can hear audio with worse quality than somebosy elsewith their sound system.

3 Likes

When there are multiple species in a recording you are IDing, politely remind/ask the observer to (in the future and possibly then, as well) include a note with the approximate time stamp of where in the recording to listen for the observation’s species.

7 Likes

There’s always going to be a level of error in observations, even research grade ones. People take bad photos, misremember the location, fabricate something entirely, etc. There’s no guarantee any single observation is going to be correct, but the hope is that with enough of them, the patterns begin to emerge. I think that anything that you can “squeeze out a fairly confident identification,” as you say, is about as good a standard for research grade status as anything else.

If you see something that you feel is really overreaching on an identification, remember you can always add your own ID and bump it back up to a more general level. For instance when there’s several lookalike species in an area and the photos just don’t show enough traits to make a specific identification. (Although it helps if you can mention why you’re doing so, and what would need to be shown in order to make a detailed ID).

7 Likes

I do not think such a line should be drawn. This is the same sort argument that comes up periodically about whether an ID should be marked as “Community Taxon still be confirmed or improved? Y/N”, where someone argues that because they can’t get to a more refined ID no one else can, which is demonstrably untrue in the majority of cases.

People very familiar with a species are often able to identify to species (or even subspecies) from a minimum of information, and the relevant information may not even be the photo (or sound) of the organism itself, but the context (eg. vegetation, location, other species heard in the audio, etc) of the observation.

There was a great example of this a few years back where there was a big argument over whether a bird pelvis found on the beach was identifiable, with one group vocally arguing that it was impossible to get it to species, and probably even to genus, therefore IDs should be closed for it. Others argued against that, and eventually a specialist found the observation and was able to immediately ID it to species based on their familiarity with the species and area.

It’s good to know your own limits, such as that you can’t make an ID from a blurry or obscured photo, but it’s not a good idea to impose your own limits on others who may have different limits on their knowledge.

6 Likes

I was talking to my friend about audio recordings on iNat just the other day. Folk seem to upload long recordings and include descriptions of which parts of the recording they are focusing on. We think that because iNat offers in app recording without providing any ability to trim or QC whatsoever (on iOS at least it seems) it encourages messy sound recordings. So I reckon this is something iNat should work on to encourage good recordings rather than deeming observations as too low quality. Plus it can be tough anyway and it isn’t fair to discourage users for providing something that is perfectly usable.

6 Likes

Yesterday there was an obs for a snake. ‘It was here, but disappeared into that plant’

2 IDs for the snake. Then a third for the ‘plant in photo’ pushing it back to languish in Life. I suggested that the plant IDer rather used DQA and said no evidence of organism.

I scroll on by myself …

3 Likes

if iNat offered the ability to clip sounds and crop photos, or else annotate sounds or photos with that sort of thing, i think it would be realy helpful. However i echo others here otherwise that no, there should be no criteria for evidence quality other than whether it is diagnostic or not, and also yes, just because you can not tell it apart doesn’t mean others can’t. Unless you are an expert who is completely certain that the photo is not diagnostic, like blurry pictures of a spider identified to species level when genital dissection is necessary to do so, you should not be adding coarse disagreeing IDs. I see this often on blurry photos of plants that are actually very possible to identify if one knows the area and species well.

4 Likes

Observations with community ID of order, family, etc. become casual grade when DQA “taxon/ID cannot be improved” is selected. About blurry photos also bear in mind that distribution can be used to ID, for example when there’s only one species of a genus known to occur in a location.

2 Likes

Technically you can crop photos…in the Android iNat app.
Sorry, iPhone users. Guess you have to edit in phone gallery :smile:

oh, i didn’t even realize that! I used to have an android phone as my work phone but don’t any more. the reality of my work flow is, i’m not likely to crop things using the iPhone but probably would if it were an option on the app and/or website.

2 Likes

Yeah, I have an Android personal & an iPhone work.
I rarely use the iPhone because the camera only goes 5x, and on top of that the feature parity between the iNat Android App & the iNat iPhone App is very disparate (at the moment, anyway…I know they’ve announced making it the same cross-platform is a goal).

Ie, The following are all things I can do in the Android version, but not the iPhone version:

  • edit photos directly in iNat,
  • see activity on others’ observations (not just my content & news),
  • add up to 20 photos just like the website (iPhone version only allows 4 at a time),
  • the “me” screen shows my species count & IDs (not just my observations like iPhone version)
  • can change from grid to list view,
  • there is a search in my observations,
  • I can see my messages/inbox,
  • I can access guides & Missions,
  • I can edit my profile’s bio/description,
  • etc.
1 Like

Agree. Or an easy way to cancel the current recording and start a new one if you find that you’re recording 30 seconds of background noise while waiting for that bird to call again (even though it was calling incessantly right before you whipped out your phone…). Definitely something I’m advocating for in the app redesign, at least as a stopgap measure to sound editing functionality (I’m making no promises, though).

6 Likes

being able to reorder sounds (like you can reorder photos) would be nice too

1 Like

yes, this keeps happening to me too. I start recording, no bird sound, I cancel it because it’s 30 seconds long with no bird sound. Then I cancel and quickly try to restart the recording. In that time period the bird calls clearly and loudly but once i get the recording started, it only catches the very end. Then i wait for another 30 seconds… etc etc :smile:

3 Likes

:joy: LOL… yes, that’s why I use Voice Memos app for sound.

1 Like

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.