Classes uploading large quantities of poor observations

Hi. During my time in Primary school, I had to take part in an interschool Inat event. Although I managed to find some decent records, I have also noted problems including-

Inability to get proper equipment- As the school had banned the usage of mobile phones( I got excused to use my camera), we, along with most other schools, were forced to use iPads, which provided non-idable photos most of the time, especially for distant subjects. However, I do not see a easy fix in this, other than to allow students to use their own mobile phones, which generally have better image quality.

Duplicate observations- As one interesting observation usually leads multiple pupils to take photos and upload them to Inat, this creates an immense amount of duplicate observations

‘Casual observations’- Pupils tend to upload most things they see, regardless if they are wild or not, as they have not been properly told about the guidelines. Therefore, most of the observations included pictures of other pupils, cultivated flowers, and even non-organic objects such as a rock. Additionally, wrong ID also seems to be part of the issue, as they usually click on the recommended IDs, without knowing that they can enter species/stakeholder names.

I believe that this problem can be solved only by the school itself, when holding the observations. As most teachers do not have knowledge about the platform nor identifying observations, it will be difficult to properly manage the observations. However, as most observations from classes usually originate from certain events hosted by the school, I believe that it is essential to set guidelines for such events and encourage the school to inform teachers about the guidelines. Maybe worth making another suggestion post for this.

6 Likes

There is a whole guide! And i totally agree if teachers did even 10 min research about iNat they could find this and prep students correctly. https://www.inaturalist.org/pages/teacher’s+guide

This actually is ok, technically (just annoying sometimes for IDers, it breaks no rules). What shouldnt happen is one person uploading the same observation many times.

8 Likes

Thank you for posting the link to the Teacher’s Guide! I was wondering if someone had already written one.

At the end it has some links to other people’s suggestions, and I’d like to single out one “Exploring iNaturalist Data in Your Classroom by Anne Lewis. Guided videos demonstrating using the species/location search, examining potential food webs in a particular area, tracking organism locations across the seasons”

I think it makes an important point about having students consult iNaturalist in their area long before they create accounts and contribute to it. (The main Teacher’s Guide says that teachers should also spend some time doing this before deciding that iNaturalist will be useful to their students.)

A school here had a project that filled up the identification queue with photographs of flowers and leaves sitting on desks, and yes, the coordinates for those observations were clustered on the school grounds! Thankfully they’ve taken those down, and some of the remaining observations are useful. But I agree that with the that, while iNaturalist can be inspiring to all ages, requiring students to use it can easily be off-putting to both the students and to passing identifiers.

8 Likes

I think this platform is good for cultivating a good hobby of observing the creatures and natural environment, so I’ll encourage students to use this app or website, regardless of whether the subject in their pictures are visible or partially hidden. I believe the way the AI works is not entirely on the picture but some big data of the place, date, and other factors. so even if the picture of observation is not clear, there is data recorded. I believe University students should be able to take good pictures. A few high school students like to take human pictures. Primary school students may not understand how the camera works, like why the birds are so far away after viewing the pictures at home or why the insects are blurred. I also get a lot of blurred pictures. so I won’t fault them in any way. The world needs more future naturalists.

4 Likes

Yes, I discovered this power tip after building many URLs by hand. It’s a great time saver!

In fact, I end up doing this all the time! Wouldn’t it be great if there were a “browse” link (right next to the “identify” link) that loaded the Identify interface with the appropriate filter options pre-checked? Of all the iNat interfaces to browse photos, the Identify interface is the best.

2 Likes

You might want to create bookmarks containing the commonly-used URLs and put it in your web browser toolbar or bookmarks menu for easy access.

For Chrome - https://support.google.com/chrome/answer/188842?hl=en&co=GENIE.Platform%3DDesktop

For Firefox - https://support.mozilla.org/en-US/kb/bookmarks-firefox

For Safari - https://support.apple.com/en-nz/guide/safari/ibrw1039/mac

3 Likes

I recently went through a Master Naturalist certification course – full of adults specifically trying to become naturalists(!!!) – and we had a presentation on iNat, reviewing the guidelines, how to use it, watching video examples of how to observe things, etc. The end result of the observations was exactly the same as described here.

IMO the only true solution would be for the teacher/instructor to go through the observations with each student, on their device, and help them through adding identifications. Even then, for quite a few people it just does not seem to “click.” You can lead a horse to water…

That being said, for some people it absolutely does click and they are off and running. To me, all the questionable observations are worth it for that.

19 Likes

Oh, yes, as a teacher of college students and high school students, I promise you that you can provide detailed lists of what to do and what not to do and then walk the students through the lists and explain in detail and re-emphasize what not to do. You can tell anecdotes about students who have failed to follow guidelines; you can set your instructions to music; you can provide an interpretative dance about avoiding pitfalls; you can paint an allegorical representation of what not to do; you can compose an epic poem with the laws embedded, and then . . . 10 students do the “what not to do.” When asked why, they look surprised and say, “They didn’t know. No one told them.” Teaching.

27 Likes

Indeed. At our university BioBlitz, we set the project to only count verifiable observations (RG, Needs ID) and tried to explain what this means and how to make qualifying observations through a variety of info channels - faculty/class meetings, BioBlitz website, iNat project journal, YouTube videos, in person training and guided observation outings etc. I still got my fair share of emails from students reporting “bugs” with the website because their casual observations weren’t counted in the project stats. The students who cared were the ones who could earn extra credit for some class, so their interest appeared to be mostly grade-motivated. The reason they knew there were issues was because several people were working very hard behind the scenes to curate the incoming observations and mark things cultivated as soon as possible. At some point early in the BioBlitz, about half of all observations were cultivated plants, but that percentage dropped later on as students figured out that those didn’t count in their stats.

11 Likes

Computer Vision offers a suggestion, but the ID comes from a person. The named identifier says this picture is That species
CV doesn’t add an ID. Identifiers bring time and effort. And we can choose to skip obs that are tricksy.
And also to support newbies who make it clear that they are interested. Have questions. Respond to comments.

1 Like

Same, but AL not LA, and they had a short presentation about iNat, and encouraged people to use it on the field days to help with how to use it firsthand. No one has been posting anything weird or ‘wrong’ though - so I’m not sure why you said that observations still same as being complained in this thread. Did I read that wrong, or were there really issues? I regularly work on our state’s observations and every single one from others in the program are excellent examples. Maybe shorter is better, plus that literally it’s worked through in person? because yes, ours, we worked through it for anyone struggling to understand on their device. It was a prereq to come with the ap downloaded already on your smartphone if you had one and wanted to learn to use iNat.

5 Likes

Exactly. We should think of this as a small price to pay to get people involved and hopefully hang on to a few that will continue to use the platform in a productive way.

4 Likes

I do that of course but most of those bookmarks are links to the Explore tool, not the Identify tool. In part, this is because it’s so easy to go from the former to the latter.

If you click the “identify” link in the Explore tool, presumably you’re in the Explore frame-of-mind, not the Identify frame-of-mind. Generally, moving from one frame-of-mind to the other can’t be anticipated, so bookmarking a link to the Identify tool is not a practical solution.

That’s right up there with the student who was absent for most class sessions, didn’t take the quizzes during those sessions (obviously), and then emails the TA two-thirds of the way through the semester with “My concern is that it is affecting my grade.” When said TA carefully went over the syllabus on day one and explained exactly how it would affect their grade.

5 Likes

Well, they’re young, and some of them aren’t really interested in education yet. My students always think I like the students who earn the highest grades, but I like the students who love to learn.

7 Likes

Primary school students may not understand how the camera works, like why the birds are so far away after viewing the pictures at home or why the insects are blurred. I also get a lot of blurred pictures. so I won’t fault them in any way. The world needs more future naturalists.

This is one of the reasons why the separate Seek app exists. Children and other casual users can point the computer vision model at whatever they like, including each other, without creating public observations. I don’t fault primary school students who use iNaturalist without understanding it. I do fault teachers who get their students using iNaturalist without properly training them, and without curating the observations afterwards.

I agree with other comments here that no amount of lecturing will pre-emptively avoid every possible issue. Teachers must be ready to spend some time with individual students going through the process for photographing, uploading and adding a suggested identification.

7 Likes

I want to make sure this topic does not devolve into consisting mostly of complaints about students being made to use iNat and posting poor observations (which is definitely frustrating!). But it’s really easy to vent on a forum (which, again, I totally understand) and that can build on itself and become pretty messy.

I appreciate the shared experiences by teachers and others who’ve encounteerd similar issues even with adults in similar classes. I think anyone who’s taught knows that not every student will understand and apply everything they’ve been told (I certainly didn’t when I was in school). What’s frustrating (at least to me) is not being able to find the teacher(s) and help them curate their students’ observations and stem the tide. I should note that when people make accounts (at least in English) the welcome email contains a prominent link to the Teacher’s Guide:

But many (most?) people don’t read welcome emails, unfortunately.

It’s a way to avoid a flood of these observations if you’re looking at the newest observations being uploaded, which is the default for Identify. Nowhere did I say it was a complete and perfect solution, but if you’re frustrated it’s filter that currently exists and can remove most of these sorts of observations from your Identify feed. Conversely it’s a way for people to find those new users and welcome them to iNat (which is usually how I use it).

9 Likes

So perhaps there needs to be a pre-defined iNat feature to allow a structure such as a class to engage, with an accountable adult at the top of the chain? Thereby, irresponsible teachers can’t just turn up and let their students loose with gay abandon. A curation filter over the top to ensure quality of obs would solve this?

Yeah I know, add a million lines of code :laughing:

I guess I’m rarely looking at such a short time horizon, so it seemed odd.

The identifer will be the “other factors” of my earlier comment. I’m just speculating. I do not know the protocols upon which CV and machine-learning operates on.
Kids these days are having to use very complex devices. I didn’t get to use a computer until I was 14 years old and there was no www back in 1980s in SEAsia. We were playing with Pacman only.
This modern system is capable of suggesting an Identification. Sometimes, AI is sure that the organism belongs to a particular genus, and often one or more species are suggested. The accuracy is not 100% but I’m constantly impressed. The system is capable of influencing the decisions made by humans, especially new users. New users may think that AI is so good that it can’t be wrong. AI may have some supercomputers with Big data centers holding enormous cache of information. Like even Chess master Garry Kasparov lost to Deep Blue.
What I think is if a picture holding limited info comes along, the AI can’t be sure but there will always be some suggestions. The user thinks AI is always right and takes up the most convenient suggestion. Afterall the community will attempt to verify that image later. The human identifiers somewhere else get frustrated by an object they can’t ID. or suspect that the AI is misled. Even for clear images, AI don’t always get it right. That’s true . Everyone knows this, partly because of the way the creatures contort themselves.
By the way, new users like to post several different organisms in one observation. By preventing this, the system will probably prevent 3-5% of errors.