0
Geographic origin of images

Am I correct in assuming that the experts being asked to make the identifications were not given the geographic origin of the images (presumably because this was not known to the study authors either, given that the images came from Google Images)?

waiting for moderation
1 Answer
0
Accepted answer

Hi Andrew, thanks for your question. We did not reveal the location, or give any clue to where the species may be found (e.g. UK, Europe etc.). This was to avoid potential bias associated with prior knowledge of the species, and for identification to be made on visual cues only. Cheers, Gail.

waiting for moderation
0

Thanks Gail, that makes sense. I'm wondering whether anyone has compared the difference in image ID speed, accuracy, and confidence with and without geographic information. From personal experience, I can normally identify most common species of snakes (my area of expertise) given a photo and a location, but (as you found) it's much more difficult to commit to a positive ID without knowing at least the continent.

-
waiting for moderation
0

I think that would be great to look at! There is literature re accuracy & speed in face recognition, but as there is limited work in species id area, I wanted to look at some basics and then hopefully build on the findings. I think that there a lot of permutations - different species, knowledge bases, experience, ages etc., so it would be lovely to see the outcomes in these different areas. (PS - if measuring speed, probably best to do in controlled conditions as this task attracted attention and subsequently some distractions!)

-
waiting for moderation
0

Thanks again Gail, I agree. We'll be pursuing this idea in the context of crowd-sourced snake identification soon, I hope. Maybe you'll be interested in this paper by my colleague Colin Shea, who showed that even experienced observers have high misidentification rates for some tricky species.

-
waiting for moderation
0

Hi Andrew

I referred to that paper in my research! Also Phil Culverhouse’s work. I really appreciate you sending me the link though. The snake id sound very exciting. I did start looking at adder head patterns (widely used for individual id), but read a paper to say that colour and scale pattern can change over time & I never got around to revisiting it! Work on automatic id for plants pays a lot of attention to shape, and I wonder if this would have any merit for photos of snake heads from above (I did geometric morphometrics at MSc and that's always in the back of my mind).

I look forward to the outcome of your snake id work, and please do not hesitate to get in touch if you have any questions/discussion points.

Thanks again for your interest,

Gail.

-
waiting for moderation