Workflow question: Geotagging photos with accuracy data included

[bouteloua]

Does anyone have an iNat workflow that produces the following?

• Use DSLR or other non-phone camera for taking photos. Camera does not have built-in GPS.
  
• Use phone or handheld GPS unit to record GPS trail of where you are walking.
• Phone GPS app/handheld unit records horizontal accuracy radius (in meters?).
  
• Computer software geotags the photos by adding both the GPS coordinates *and the accuracy radius* to the photo metadata.
  

Or maybe there's some even better way of doing this?

Without a) having to take points manually with the phone or

b) buying a separate GPS receiver that connects directly to the camera.

grazie,

cassi

[Murray Dawson]

This is what I do, but of course there are numerous other approaches from those more expert than I.

Even if your camera does have a GPS, sometimes you still need to correct the location if you have taken the shot too quickly or if it is too inaccurate.

I use Picasa to optimize my images, add captions and edit my geolocation data - Picasa has an easy interface for doing this. Unfortunately Picasa is now unsupported and a registry fix is needed for the geotagging to work (https://productforums.google.com/forum/?utm_medium=email&utm_source=footer#!msg/picasa/mHZcTDkUrMQ/UKRLS-YMDAAJ). Alternative free software that edits geolocation data is the multiplatform Digikam but it's more pro grade and I haven't used it. Don't know if it sets an accuracy radius.

Once my images and metadata are tidy I upload them to Flickr as a repository which gives 1TB of free storage and protects me against local hard-drive failure. I make all images there private just to me, but share them through the iNat feed. Choosing "add metadata" in iNat pulls across the image captions so I don't have to write them again.

This does not answer all your questions but I'm sure useful comments from others will follow.

[bouteloua]

I can already link the GPS track coordinates to all the photos I take in an outing with a few clicks via Lightroom. But Lightroom does not seem to include any accuracy data.

But thank you, nice to see the different ways people are going about adding observations!

cassi

[Andy Kleinhesselink]

Cassi, 

This a great question.  I'm working on this right now after I realized a lot of my obs. don't have accuracy information.  

There's two issues here that I can see, first is assigning the gps accuracy data to the photo metadata.  Second is getting the accuracy data into iNaturalist.  I can help with the first issue.  But I don't believe iNat will read in accuracy data from the photo metadata (you probably knew this already):   https://www.inaturalist.org/posts/1869-the-need-for-assigning-accuracy

I believe the standard exif tag for accuracy is "GPSHPositioningError"  <https://www.sno.phy.queensu.ca/~phil/exiftool/TagNames/GPS.html>

So basically you want to edit this tag for a bunch of photos. 

Assuming you know the accuracy of your gps, you can use the command line program exiftool. Exiftool is a free command line program for editing photo metadata.  If you know the accuracy of the gps track, you can use exiftool to quickly edit the "GPSHPositioningError" tag.  If you are on the command line and have exiftool installed this would be the basic command:  

exiftool -gpshpositioningerror=10 example_photo.jpg  

This would assign a gps error of 10 m to the photo named "example_photo.jpg".  

You can edit this tag for every photo within a directory at once by leaving the file name of the photo out and replacing with a period.  

exiftool -gpshpositioningerror=10 .  

So that would be an efficient way to do this for a lot of photos at once assuming they all had the same accuracy.

A separate issue would be if you have a track file that includes gps accuracy data.  In that case, I think the exiftool -geotag function would automatically add the accuracy information if the gps track file had that information to begin with:  https://sno.phy.queensu.ca/~phil/exiftool/geotag.html

However, I haven't been able to try this because my gps track files (from a garmin gps watch and from the openGPS tracker app on android) don't include the "GPSHPositioningError" tag.  

So none of the above is a very complete solution I'm afraid. 

Also discussed here: https://groups.google.com/forum/#!searchin/inaturalist/gps$20accuracy$20%7Csort:date/inaturalist/D7FG2MzRsEU/rAx8xNRhBQAJ

Let me know if that is helpful, 

Andy 

[bouteloua]

Thanks for all of this Andy! I think I would be happy enough assigning a very rough accuracy to all the images in a batch, then fine-grading them on a case-by-case basis as desired.

So, I tried out exiftool.
I assigned a GPSHPositioningError of 10 per your previous post.
I checked the exif data to make sure it had written correctly - says "10 m".
I then tried uploading it via the iNat browser uploader (https://www.inaturalist.org/observations/upload).
Looks like the coordinates come through but the accuracy radius still does not get assigned.







I attached an example written photo. What am I doing wrong? :)

cassi

[megatherium]

I've been bypassing EXIF and just bulk-assigning uncertainty in the uploader.  Would that not suit?

--
You received this message because you are subscribed to the Google Groups "iNaturalist" group.
To unsubscribe from this group and stop receiving emails from it, send an email to inaturalist+unsubscribe@googlegroups.com.
To post to this group, send email to inatu...@googlegroups.com.
Visit this group at https://groups.google.com/group/inaturalist.
For more options, visit https://groups.google.com/d/optout.

[cassi saari]

Ooh, I had assumed batch editing any GPS info would overwrite the individual coordinates. Looks like that's not the case and that accuracy is updated separately.

In lieu of some mythical phone GPS tracking app that logs "true" GPS H accuracy, looks like this would work.

thanks!

cassi

[Andy Kleinhesselink]

Cassi, I had the same experience.  (sorry if my response was misleading). The exif metadata for accuracy does not get read in by iNaturalist.  So assigning accuracy via Exif is currently not a clean way to get it into iNaturalist.  

However, I think the developers could probably change their desktop photo upload tool to read in the gps location and the accuracy metadata.  The gpshpositioningerror tag is fairly new if I'm not mistaken so it maybe that they didn't include it because it wasn't available a few years ago.  

I've seen some threads and disucussion arguing that this is why you should use the app to input observations rather than uploadloading photos separately.  But that just doesn't seem tractable for most taxa--it's just really hard to get a good telephoto photo or a a good macro photo with a smartphone!

Consider this response by Ken-Ichi:  

"Personally, if I think I can capture my subject in a single shot, I 

just take a picture in the iNat app. If I think I need to take 
multiple shots and edit them (multiple angles, moving bug, etc), I'll 
take my pictures in another camera app since that makes it quicker and 
easier to focus on getting the pictures I want, then I'll immediately 
share the pics I like from Google Photos to the iNat app and re-fetch 
the location to ensure I get GPS accuracy. For maximum efficiency I'll 
just use my SLR (no GPS accuracy, but much faster and sharper). "

https://groups.google.com/forum/#!searchin/inaturalist/gps$20accuracy$20%7Csort:date/inaturalist/D7FG2MzRsEU/rAx8xNRhBQAJ

I don't think I understand what he's doing there. Maybe I'm missing something but how do you use the app with a separate camera? 

I'd advocate including accuracy metadata (where available) in future versions of the iNat uploader tool. 

For what it's worth the openGPS tacker app for android records accuracy along with track.  It's a good tool for geotagging photos, but exiftool doesn't seem to use this accuracy tag.  Might need to write my own script! 

Andy 

[Tony Iwane]

I've made a github issue for supporting GPSHPositioningError in the web uploader: https://github.com/inaturalist/inaturalist/issues/1684

Tony

[Mark Rosenstein]

This issue has come up repeatedly over the years.  Because accuracy information isn't very important to me when most of what I observe is flying/swimming organisms who cover a lot of distance, and my tool chain does not support tracking this information, I don't provide it in my observations.  However, I do see the value in it.  I would like to see support for this added to the tool chain used by many nature photographers.

Can someone who understands the issue better than I write up something that can be the basis of requests to Canon, Nikon, Adobe, and other such companies to support this information throughout photo processing?  If enough of us request it, in a few years this might just work without any manual effort on the part of those of us using real cameras rather than smart phones.

-Mark

[Cullen Hanks]

It would be awesome if we could more systematically assign accuracy based on the GPS.

For now, the batch loader makes it really easy to assign accuracy to all your observations at the same time.  Just select all, click location, and you can add an accuracy to all your observations.  This assumes they all have the same accuracy.  However, a conservative estimate of accuracy is way better than no accuracy.

-Cullen

--
You received this message because you are subscribed to the Google Groups "iNaturalist" group.

To unsubscribe from this group and stop receiving emails from it, send an email to inaturalist...@googlegroups.com.

[car...@inaturalist.org]

Update: the observations uploader should now automatically incorporate GPSHPositioningError if it exists in your photo metadata, thanks to Ken-ichi. 

Best,

Carrie

[Andy Kleinhesselink]

Thanks Ken-ichi! I'll try this out soon.  

[Chris Vynbos]

What is best practice when one can't get an accuracy reading out of one's gps track or EXIF? Does one leave it as 'not recorded' or does one add a guesstimate? 

Andy said "Assuming you know the accuracy of your gps " how would one know that? At times my track can be way off the route of my hike, so one assumes one's images have wrong gps data too.

[bouteloua]

When I record a track with my phone, I find that when reviewing the datapoints afterwards, they usually seem pretty spot-on. It's often when you take one-off images with the camera app and don't let the GPS connect to the satellites for long enough that you get some crazy inaccurate readings.

I don't think there is any sort of established best practice, but if you can review the observations for any potential stray data, I would say that 10 meters is a pretty reasonable guesstimate for phone GPS when recording a track in open conditions. If it's a really poor signal situation like heavy tree cover, canyons, skyscrapers, etc, you'd want to bump the accuracy radius way up and possibly manually change the location if you know it was totally off. I don't think there's any reason to leave the accuracy radius unentered, besides that it may take a little extra time for review. If you *really* had no idea where you were and if you also suspect your GPS device was recording inaccurate data, you could (should?) make the radius super huge to relay that information to others.

Best case GPS scenario, rarely achieved, for most current smartphones is an accuracy of around 3 meters I think. If you're using a separate GPS device you can probably check the specs or request info from the manufacturer.

cassi

[Chris Vynbos]

Thanks. I was hoping that there was an app that gave constant accuracy readings while recording a track based on how many satellites it was able to 'see' at any one time.

[Tony Iwane]

I have yet to find a way to record horizontal accuracy when keeping a track while in the field and then geotagging later, and I've tried several types of software.

I try to use my Garmin eTrex 20 when I'm in the field and when I go to the Satellite option on the device it usually shows between 2-4m accuracy unless I'm in a narrow canyon or under a lot of cover. It's a good little device because it uses both U.S. and Russian satellites. And I agree with cassi that my locations always seem very accurate when I look at them on the map. Taking a quick photo with your smartphone is likely to be much less accurate. Fortunately iOS devices now record horizontal accuracy when taking photos, but I don't believe Android does...

Tony Iwane

[Chris Vynbos]

I record my track with my smartphone (samsung), and when i review the track on google earth it is usually right on the path visible on google earth, so no problems there, except as you say, it goes AWOL when in deep gorges. Even if you have a garmin, you might (if your phone battery allows it) find a benefit in recording a track with your phone at the same time as it means your phone's camera can add accurate locations to the photos with no delay. 

[Charlie Hohn]

yeah, i take lots of observations and usually when i have an error, it's just a couple of observations and they are clearly off the track. So I move them on the track with a larger uncertainty circle based on where I know i was (you can use timestamps too). Older versions of iNaturalist and/or older phone OS used to have bigger problems, like observations popping up way off of the line, but I haven't seen much of that lately. I did notice that the android app was slow to zone in to locations yesterday but i was on a slope near a mountain ridge so that is probably why. 

Also, like cassi i've also noticed that if you take a 'one-off' photo the location often doesn't record well.

--

You received this message because you are subscribed to the Google Groups "iNaturalist" group.
To unsubscribe from this group and stop receiving emails from it, send an email to inaturalist...@googlegroups.com.
To post to this group, send email to inatu...@googlegroups.com.
Visit this group at https://groups.google.com/group/inaturalist.
For more options, visit https://groups.google.com/d/optout.

--

============================
Charlie Hohn
Montpelier, Vermont

[Mikael Behrens]

Thanks for the interesting discussion!

I take lots of bird photos for iNaturalist with a telephoto lens on a micro four thirds camera. I use Gaia GPS to record my path, and Lightroom to geotag my photos with that path. Even if this workflow included accuracy for each point, often the bird I photographed was outside of this accuracy circle. For example, let's say the point I was standing on was recorded with 10 meter accuracy. But I photographed a bird 30 meters away. So I try to set accuracy values that include about where the bird was.

Does that make sense? Am I doing it right? :)

Mikael Behrens

Austin, TX

[cassi saari]

Sounds right to me!

I mostly photograph plants myself, so the "right spot" for the observation matches where my GPS is.

cassi

--
You received this message because you are subscribed to a topic in the Google Groups "iNaturalist" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/inaturalist/IPQEKOj3J58/unsubscribe.
To unsubscribe from this group and all its topics, send an email to inaturalist...@googlegroups.com.