Many observations lack photos of key parts of an organism that are important for identification. For example: close-up photo of a flower but no leaf photo, top of mushroom but no gill pattern, back of an insect but not the face, I’m sure there are hundreds of examples.
To help beginners or even people who are observing an unfamiliar taxon, have optional photo prompts that pop up after a taxon is selected. The prompt would say something like “These parts are important for identification” then list the parts, then have two buttons “Add more photos” and “I’m done.” Turn it on by default for new users but let users turn the feature on or off in the settings.
I think the list of parts should use very simple terms, not something that needs a glossary. For example: back of flower, upper leaves, fruits, top of head, lower leg, top fin. Sometimes habitat or host is really important so that could be a pointer too.
It won’t be enough for the most difficult taxa but I believe it would help both identifiers who want a little more information and observers who are frustrated that their observations don’t get identified.
Some administrative points:
The list should be short. Don’t make users scroll through a laundry list of desired shots. 3 or 4 should be enough.
It should probably be restricted to curators but let everyone flag the taxon to request changes. Boiling down the most important parts of a taxon takes some research and/or experience.
The curator should be able to copy the list to direct descendant taxa.
It’s probably less important for the web upload since most users are uploading photos taken much earlier.
Are you thinking that the suggested parts would be suggested based on what the AI thinks the observation is? I could see users getting confused by suggested parts that don’t make sense because they were suggested for something visually similar but unrelated.
This is mostly what I do, so while I would love automated photo prompts in the field, I’ve just accepted that I take pictures when I can and learn as I go. Even then, it might still be helpful to see when I upload so that it can tell me for next time without waiting for a user to specifically tell me what to improve.
I can see where you’re coming from, but I think having these sort of prompts in the field might do more harm than good. It’s hard enough getting a photo of, say, a butterfly before it flies away without having to fight with this sort of pop-up.
It would be more helpful to have some sort of separate guide to taking good photos, something like a couple of pages of PDF with suggestions for each main category, like photoing the gills of mushrooms and not just the top.
I wouldn’t recommend this for the camera screen, only after someone has added their photos and has selected a taxon. If someone is entering the other data, they’re probably done chasing that butterfly.
I see that as a bonus. If I take a photo and select a suggested taxon and the prompts don’t match at all (for instance legs or wings on a plant) I would realize I’ve picked a taxon that’s not even close to what I’m looking at and try again.
This is a great suggestion. Also, as it would be optional, it wouldnt need to be answered (or even available for all species). So if the butterfly flew away, the user could still submit the obs… but if an extra photo of some leaves, sepals etc could help id a plant, that would be great. And it would really teach the amateurs (speaking as an amateur who loves this app)
It’s a great idea and we’ve discussed this among staff… ad nauseum, at this point. The main problem is sourcing the content: is this something we keep incredibly simple and have staff author and control, e.g. “if taxon is Agaricomycetes inform the user they should get pics of the top of the cap, the underside of the cap, and the base of the stipe,” or is this something much deeper and crowdsourced, e.g. “if it’s a garter snake in California make sure you photograph the upper labial scales and check for ventral spots”? Both would involve some significant UI design, but the latter also incurs some serious infrastructural and moderation burden (are you ready to have a new thing to argue over?). That’s not to say we won’t / can’t do it, just that it’ll be quite tricky to do it right for everyone everywhere the planet.
I’m glad the staff likes the idea and I certainly understand the challenges of designing a whole new section of data and interface.
I don’t think using really fine anatomical details will serve the majority of iNat users very well. The vast majority of users are not professional biologists and even those of us who are, usually only specialize in one area. I think the goal shouldn’t be to get the perfect photos to key out every subspecies but instead to help non-experts get better photos that are good enough to get a pretty good ID.
For the technical details, there’s always the Wikipedia page and there will always be species that are impossible to ID from photos.
I think you could phase it in by starting with very broad anatomical terms at the class level or higher (flowers, fruits, leaves, and whole plant would probably be sufficient for 80% of vascular plants) and allow curators to create terms at lower taxonomic levels as needed. The terms should be geared toward beginners and be easy to translate.
One way to code it might be to look up these terms starting at the selected taxon and if they’re blank then working up through the taxonomic tree until a set of terms is found. That way you wouldn’t have to push changes to lower taxon records.
Making iNat “right for everyone everywhere” is a lofty goal but it puts a lot of pressure on the development team. I think it’s already pretty great and getting better all the time.
I agree the words used should be simple. Maybe have a little (i) information symbol for specialists, so that it shows what the anatomical scientifically correct name is. But mostly just show simple words.