Getting Paid to ID?

Yes, it is a favourite argument of property developers that fragmenting a block of habitat will increase the species richness thanks to the edge effect.

7 Likes

I got an image like that as well (It’s probably the same for everyone) and immediately switched my answer to “I can just do animals”.

I was also a bit skeptical about their questions, or rather categories, there, but yes, we’ll see. Haha

1 Like

It is interesting. The people photographing distant green stuff do show they are capable of taking close ups, judging by the amount of “tree bark” observations in the ‘unknown’ pile.

5 Likes

Being skeptical of the whole thing, I e-mailed Pivotal asking for more information. I got some from a Dani Rabaiotti. Sounds like a project for dumbing down botanical surveys (i.e., making them cheaper). Perhaps financially viable for all I know. Here’s the response. (The links cause potential downloads to your computer but don’t link to something on the internet; I didn’t want to download, so I didn’t see what they are.)

From Dan Rabaiotti:

As for what Pivotal is – we are a small company that measures biodiversity for land owners and land stewards. Our aim in not a comprehensive survey at the site that detects every species, but instead we aim to characterise if biodiversity as a whole has gone up or down (or stayed the same). We collect this data for a variety of clients – from NGOs looking to see general trends on their sites, to farming co-ops, to companies looking to use the data for things like TNFD disclosures, or to evidence no biodiversity loss for a number of regulatory reasons. We are also the data partner on the Plan Vivo biodiversity credits pilot. We collect digital data meaning that there is a robust audit trail to every species we find at a site – important as it means data cannot be faked or gamed. We collect data on bats, frogs and birds using audio, mammals using camera traps, and plants (which we group into low lying or ‘trees’ depending on data collection method) using drones and cameras) and combine species level data into a number of biodiversity metrics (and use a couple of habitat related ones too) that track change at a site over time. Digital data collection also makes surveys affordable and feasible for a wider range of clients who may not have the time or money to enlist a large team of ecologists to do a comprehensive survey.

As for the images in the form, they are there just as an illustration of general vegetation imagery, apologies if this was not clear from the text – we have to demonstrate that because many botanists (as you know I am sure!) need detailed close ups of a single plant to be able to do identifications, which is tough with our set up (more on that below). Unfortunately for us windows forms compresses the images horribly which makes them look much lower resolution than the are – I have attached a couple of originals for you to look at below, let me know what you think (I know they’re still pretty tough – we know plant ID using these methods is very challenging, and we are aware that for many taxonomic groups getting to species or even genus is often impossible):

https://pivotaldataproc.blob.core.windows.net/site-data-fs/Freixo_do_Meio/survey_2023/sanitised/production/camera_tripod/sample_9/place_2/DSC02264.JPG?sv=2023-01-03&st=2024-07-23T10%3A21%3A39Z&se=2024-07-24T10%3A21%3A39Z&sr=b&sp=r&sig=ZmHTp2JGvq1pBF8fOPjro%2FFakxchNKVNIp3osmSgM2k%3D

https://pivotaldataproc.blob.core.windows.net/site-data-fs/GBRFEN/00/sanitised/production/camera_tripod/sample_HQ2/placement_1/P71A9055.JPG?sv=2023-01-03&st=2024-07-23T16%3A44%3A48Z&se=2024-07-24T16%3A44%3A48Z&sr=b&sp=r&sig=qDzO7Hk3IyFfmMIQcgodrb8uXAyssO6KGbqwXIPZgj0%3D

The reason for general vegetation imagery rather than photographing individuals plants is because it allows us to photograph a far higher number of plants rapidly during data collection than it would otherwise – we know that only some will be identifiable in each image, but because we take so many, most species present will be identifiable in at least some images.

7 Likes

Anyone get far enough to know what it pays?

reminds me of https://forum.inaturalist.org/t/what-is-the-most-observations-you-have-gotten-from-a-single-photo/51157/2

3 Likes

Thank you! What do you think about their reply? Does it make you any less skeptical?

I still don’t know how I feel about this. I think the idea in principle is good. I can see the benefits of making biodiversity research/monitoring cheaper for companies. I’m just not sure it should be made by a for-profit company, which I am assuming Pivotal is(?). It sounds a bit like it has the potential to be yet another green-washing tool that allows Pivotal’s clients to claim being certified saints of environmental protection, something I’d definitely not want to be a part of.
Especially the “Plan Vivo biodiversity credits plot” has me worried, though I admittedly haven’t researched it.

Once I finish exam phase, I might do some more in-depth research on all the stuff Pivotal is linked to…

4 Likes

I haven’t, but I think Pisum’s analysis will be accurate. I think it also said somewhere that the amount paid will be agreed on with everybody individually.

I mean, first of all minimum wages differ quite a bit (12.5€ where I’m at), as do tax requirements. I wouldn’t be surprised if they would primarily aim for hiring people in countries where these costs are lower.

3 Likes

[Tag this post as humor]

:rocket: Are you tired of slogging away at a dead-end job, only to come home and do iNaturalist IDs for FREE?! What if I told you there’s a way to TURN THOSE IDENTIFICATIONS INTO CASH? :dollar: :money_mouth_face::moneybag:

Introducing our groundbreaking course: “Cash in your Passion: Get Paid to ID!”

For only $19.95, you’ll learn how to make MILLIONS by leveraging your expertise with iNaturalist! :sparkles:

:rotating_light: Just look at these results!!! :rotating_light: (Shows fake screenshots of bank account)

Yes, that’s right — thousands of dollars AUTOMATICALLY deposited into your account while you sleep!!! :sleeping:

But hurry! This incredible offer is ONLY $19.95 for the first 50 applicants! Act fast, or you’ll miss out on this opportunity! Prices go up tomorrow!!!

:point_right: Don’t waste another minute IDing for FREE! Cash in your Passion today!!! :point_left:

11 Likes

I asked Dani Rabaiotti about disturbance increasing biodiversity. Response:

"Definitely! We exclude invasives from all our metrics, and additionally that’s why we tested our metrics carefully on these kinds of scenarios and chose ones that best track change towards a healthy ecosystem, rather than simply going up if there’s a proliferation of disturbance loving species. If a handful of species dominate then that will have a negative impact.

“I saw this myself first hand many times working in Madagascar – post burn there was just an explosion of invasives – not the sort of thing you want to incentivise!”

8 Likes

I would be interested to know what their metrics are, if such information is available, especially if they have published a report or description somewhere.

I am not any kind of expert, so maybe people with more skill and experience would feel differently, but personally, looking through these photos sounds like a frustrating slog rather than a fun challenge. They mention they are aware that not everyone can ID from these photos, but they don’t bring up that people probably also don’t want to!

Maybe I’m completely off-base, but it seems to me that they would have to pay pretty well to expect people with any real level of expertise to do that.

4 Likes

Very true.

I wonder whether @pdfuenteb, the prolific iNat identifier named in that profile, would be willing to share any thoughts on the experience of working as an identifier for Pivotal?

4 Likes

2 posts were split to a new topic: Have the rules on job postings on the iNat Forum changed?

Last year I spent several weeks reviewing about 300,000 camera trap images that had been loaded into an Access database and added species IDs and other annotations to them. I was paid for that but it was part of my full time job. It was hard on the eyes after 8 hours each day.

6 Likes

that’s a lot. if i assume 40 hours per week for 10 weeks, that’s still 750 images per hour (12.5 images per minute, or <5 sec per image).

1 Like

Depending on how the camera traps took pics, a lot of the pics can have nothing in them (like they are just taking a pic a min, etc.). I’ve worked with students on processing camera trap data, and they are often just arrowing very quickly through pics (1/sec or maybe faster) until something appears in the frame and then marking that pic to come back to later. They then ID/code a much smaller subset of pics.

I think there are AI programs that can screen pics down to ones that likely have targets of interest in them for manual review now though.

My best day was 18,000 images/day reviewed. An average day was more like 8,000. But to be fair, most were large mammals so the review time per image was pretty short, a few seconds.

4 Likes

I’ve run multiple camera traps on cliff faces with crevices lizards used for places to hibernate for over a decade now. My time-lapse cameras (can’t use motion sensitive since the small lizards don’t reliably trip the IR sensors) can each produce ~6000 images a winter and I can score those (by counting the lizards visible) in a couple hours (here, here, and here are a couple recent favorites). On an unchanging background, things blink in and out and deserve attention, but not all images do and it becomes toggling back and forth to see what has changed between images. The key is to have as little movement as possible in the focal zone to minimize false positives. It’s actually quite similar to IDing on iNat. Since I just (yesterday) finished scoring 2023-24 data, I’ll share what it looks like (even though it’s off topic…it’s just that cool). Lizards come to a particular crevice in the fall and create a peak date on which they are visible in photos, then winter sets in and they go mostly inactive waiting for those favorable days. In the spring more and more emerge and eventually leave the field of view to set up territories. So each colored dot is a camera trap image scored for non-zero lizard days (the y-axis doesn’t include zero so that any dot shown is positive for a lizard day and stacked columns on the same date indicate different hours activity). The pattern seen here is the pattern every year and after a decade of this the data suggest winter is getting shorter by about a day a year, mostly because lizards are able to be active later in the fall (late November in the data shown).

5 Likes