Decided to give Seek a try on a day I felt like spending 2 hours to identify 45 species of plants around the neighborhood. Was impressed with all of the taxonomy crammed into its offline AI. Came home to connect to Internet and upload observations to iNaturalist, only to discover that Seek wiped geotags from all of the photos. Don’t feel like using Seek ever again if that’s not fixed. Bummer.
Can’t decide if this constitutes a bug report, a feature request or if I’m just doing something wrong. Please advise.
So, erm, 5 years later it’s still unresolved and now also creeped into the Android app?
And the only reason cited is that EXIF metadata writing isn’t implemented in Seek…? But I can see that camera settings metadata is embedded into the files, so the app does have the capability to write EXIF metadata.
Yet the developers deliberately intend for me to go through the rigamarole of capturing shots with another camera app and then importing them into Seek, even though the Seek app already tracks GPS…?
That’s the intended way of use now if I want the GPS data to not be scrubbed…? Wanna make extra sure I’m not missing a setting or something sensible unlike that route.
The linked thread does say Seek was never intended to be used with iNat, and frankly I don’t use it at all. It somehow manages to make my slow, difficult phone camera even laggier and more blurry. I just use my camera, share to iNat and cut to the chase.
Seek might be fun in the field with the kids as a standalone activity, but it’s an obstacle to interacting with iNaturalist. I don’t know that it’s a priority to make it be otherwise.
Well, okay, let me highlight the benefits of Seek as I see them.
As someone who doesn’t connect to the Internet outside of home, I find offline identification at a glance to be rather useful. It also helps with checking if a plant I find is a new species or a duplicate of one I already observed before but looking different for whatever reason. And the dynamic mode of the Seek camera helps align the shots in a way that allows more confident identification from the resulting photo, which I, as an amateur naturalist, might not always have the best sense for. I take it that if the AI is able to identify the species from a certain angle — it’s probably a projection that includes all the important visual traits.
Correct me if I’m wrong, but that whole dynamic AI camera isn’t part of iNaturalist app, right…?
I have always assumed that Seek was INTENDED for children. That it does not keep much data is intended for that reason. It’s nice that parents and teachers can safely introduce children to nature and the beginning of learning identification, without concern of tracing the children’s whereabouts. It can be a scary world.
I use only my iPhone for photos and iNat. I don’t have internet at home. All my “computer work” IS from iPhone.
I use Seek for a quick id, but I have found that using my clip on lens with the Seek app can really help the closeups…
i went back and reread the post i referenced before and did some testing on my phone, and i think the way to interpret the post is that since version 2 of Seek, neither iOS nor Android Seek would store location in EXIF.
i haven’t tried the version of the iNat app that’s under development, but based on the code and the comments that I’ve seen from folks who have used the app, there should be some sort of camera mode that i assume is similar to what Seek does. it sounds like that app is being beta tested now. so i would expect that to be available within a year, assuming no big surprises found during testing.
it’s unclear whether the new version of this camera would capture location in the image file EXIF though. if you wanted to test it out, you could look at the GitHub repo for the iNaturalist React App, and install it yourself.
i don’t think so. after you take a photo in Seek, if you click the button to post to iNaturalist and your location services are on, then you will get a location on the iNaturalist observation. however, if your location services are off, then you get no location and have to select one from the map.
I’ve heard that Seek comes across as more confident in providing IDs (because it is set up in a gamified way), and I’m pretty sure that Seek is also using an older version of the CV. If you are looking for accurate IDs, you should use iNaturalist instead.
This is the kind of thing you get the feel for as you go! But best by taking pictures freely – also lets you get nicer compositions that the Seek AI might not recognize as ‘good for IDs’. You can’t get any better by relying on something besides yourself each time in my opinion.
If you are looking for accurate IDs, you should use iNaturalist instead.
I believe the issue is that iNat neither (1) can automatically ID while offline nor (2) allows you to see a provisional ID before going through the hassle of taking a picture or pictures and sharing them to the app. Seek does both of these things, and I agree with OP that it’s useful for that.
For today, however, I decided to enroll into the beta of the Play Store iNaturalist app, which didn’t seem any different… And then I went around using Seek again but this time pressing the “upload to iNaturalist” button right after capture, since that is the only time you can opt to do so. Turns out it caches the submissions offline to push them once connected. That worked out reasonably well, even though you don’t get a good preview of the photo in case it turned out blurry and can’t edit the annotations — still made submissions that I was able to curate and edit in iNaturalist once I was home.
@upupa-epops, actually, there is no such setting to automatically submit. There is an ability to log in which enables manual prompting of submissions after taking a shot and there is an option to take shots automatically once the AI recognizes the species, as a safety against the headache of it losing confidence by the time you thumb the shutter button. Also, geotags are in fact included in this specific workflow.
I have to say, this is an awfully rigid interface, but it does most of what’s needed of it if you do precisely what it wants.
Installed iNat Next using Obtanium to test tomorrow. On a glance, looks sleek, has the AI camera, laggy and crashy. As expected.
