Maximum number of observations per upload-session

Is there a maximum number of observation per upload-session? I’ve normaly doing 100 - 200 observations per day during the weekend. Using the browser to bulk upload them I’m often facing the problem that the CV doesn’t provide me a suggestion (or takes extremly long) at a certain point in this process…

2 Likes

I don’t know of a hard cap (though there might be one), but performance definitely declines as the number of concurrent observations/photos gets high. Staff has previously recommended working in smaller batches to avoid this.

4 Likes

The map is slow or unresponsive too while photos are loaded.
I wait until the number of photos shows up under each observation, the performance returns to normal then.
I batch them up by location, also running Task Manager to keep an eye on memory. If memory runs out, everything becomes slow or the laptop crashes.

1 Like

actually, i uploaded maybe fifty yesterday and i thought the vision AI was struggling more than it usually does. possible context window issue, is it preloading something that trips it up with a lot of uploaded data? i may just be imagining it though.

When I do bulk uploads, I only upload about 30 obs at a time. That is mostly because I’m scared of doing a bulk upload over 100 obs, and while I’m adding CV ids on the 98th obs, the internet crashes. Or a cat unplugs the desktop…

5 Likes

I cap it at about 50 personally. After that it seems to get a bit unwieldy.

3 Likes

Personally I drop around ~70 photos into the upload field at once (those generally get consolidated into fewer obs). I find if I go much higher it starts to chug or get stuck.

1 Like

Do you mean stuggling as in was slow to provide answers, or struggling as in was providing worse answers than you expected?

1 Like

they seemed more genus level than usual exact taxon. im still learning what a good picture is in terms of getting an ID, possibly even the light was what affected it noticeably

No official cap as far as I am aware but your system memory, your choice of browser and your internet all seem to have an effect on that “cap”
For me on my laptop on my home wifi anything over 25 observation with 1-5 pictures each will take exponentially more per additional observation.
When on a more powerful desktop with an ethernet connection 200-300 seems to be the slowdown point.

I have also seen this where the cv is less accurate/lower certainty on the upload page

I tend to not put more than 50 photos in any one session, and that’s what we recommend. You can put more, but if anything happens then you lose that much more work.

As for the CV times struggling, it might just be due to higher load right now as northern hemisphere spring ramps up and the City Nature Challenge starts. My understanding is that the upload page is much more client-heavy, meaning it’s using more of your computer’s RAM and such, than it is on our servers. But I may be wrong.

Without specific examples it’s difficult know for sure, but on the observation page, if the observation is already identified as something that is in an iconic taxon, the CV will restrict results to that iconic taxon.

1 Like

Some identifiers have recommended this, too. It’s mighty annoying, when identifying by location, to come upon several pages of all one person’s observations of a taxonomic group I know nothing about. Let other people have a turn.

1 Like

Sometimes I take 500+ photos on two cameras a day.
Preparing photos, loading them as observations and updating locations takes a predictable time.

The limiting factor of the batch size for me is the time it takes to identify each, which varies from instant recognition to a full night of research.

1 Like