-
-
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
-
The classic! We'll have some similar challenges (with 4K recordings)
Kraj razgovora
Novi razgovor -
-
-
So many questions, but to start: 1. How did you teach your
#neuralnetwork to auto ID? Cronin et al. (https://doi.org/10.1016/j.jbiomech.2019.02.021 …) used a stationary camera, but I want to create a network that does what you did (e.g. ID landmarks on a video taken with a camera that isn't on a tripod) -
Frames are taken separately (e.g. there is no time relation between them). For the network it doesn't really matter if the camera is stationary or not, it evaluates them individually :)
Kraj razgovora
Novi razgovor -
-
-
Is this automatic?
-
Indeed!
Kraj razgovora
Novi razgovor -
-
-
Glad to see this working! How does it perform on variable distance?
-
Seems to work pretty well as well, but we trained another neural network for the wide lenses, to be on the safe side haha
Kraj razgovora
Novi razgovor -
-
-
Won’t these devices put them at greater risk of predation?
-
The colored dots are only digital! No animals have any markers on them - just digitally in the videos. :) Check out more
@DeepLabCut &http://www.mousemotorlab.org/deeplabcut - Još 1 odgovor
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.