New blog post is up!
-
by mason_UWI scientist
This slipped past me it seems, but a new blog post is up!
The figures look to need some updating, but it's about all of you!
Cheers,
Mason
Posted
-
by DZM admin
Awesome. Thanks, Mason!!
Posted
-
by escholzia
I have to admit I'm ambivalent about being one of the top classifiers 😃. In addition to being an interesting project, it's a great stress reliever after work. But this makes it only too clear how much time I spend on this. It gets kind of addictive. I know you won't like the comparison, but it's a bit like playing a video game .... spot the animals...... with the added benefit that I learn something now and then.
Posted
-
by mason_UWI scientist
To be honest, it's a bit of a game to me as well! I find myself saying "Just one more coyote" most days.
Posted
-
Me too 😃) or even just one more mammal!
Posted
-
by escholzia
This raises some statistical issues. If one person classifies 100,000 pictures out of 600,000? (1.7 million classifications/3 classifications per picture?), that's about 17% of all the pictures. How many classifications would it take before you see systematic error due to one person's classification tendencies? Asking for a friend.....
Posted
-
by mason_UWI scientist
There are a lot of variables that could potentially introduce bias, but just because one person classifies a lot of photo does not mean that bias will inherently be introduced. What if, for example, they were good at identifying the photos? Additionally, it still takes more than 1 person to 'retire' a photo, so regardless of how many photos one person classifies other folks still get a vote on it. However, you are correct in assuming that a users probability of correctly classifying a photo are not identical (it probably follows some distribution), and there are ways to properly account for this statistically (beta-binomial models are a good example).
Mason
Posted