Want to conveniently explore USV data generated in @DeepSqueak_USV? Check out SqueakR: a free, open-source experiment superstructure created by @SimonOgundare
SqueakR also has an #rshiny dashboard to create visualizations without any code!
https://t.co/t8igjkoYcZ
Elizabeth Ferguson chats with @NPR regarding our use of DeepSqueak, a tool originally developed to detect rat vocalizations, for detection of marine mammal calls.
#oceanscience#deeplearning#passiveacoustics
https://t.co/2tY0CVmxvq
Happy to announce a long-awaited exhibit from @ian__ingram, an LA-based artist who builds robots that interact with animals. He turned his attention to rats for this show @UCIrvine, much to my delight. Come by the opening this weekend if you're in SoCal!
https://t.co/779fYxXYML
Detailed instructions for using DeepSqueak v3 can be found in the Github Wiki:
https://t.co/0QibR9XEQw
General questions and discussion are happening on Gitter:
https://t.co/yy2GXjacPZ
Introducing DeepSqueak v3!
This is no minor update; this is a complete redesign based on 2 years of feedback from the USV community.
Explore the improvements and new features below.
Big thanks to Russell Marx, Robert Ciszek, & Lara Valderrabano!
https://t.co/pnkFkkOHRe
We have improved contour-based clustering and added NEW contour-invariant clustering with Variational Auto Encoders. VAEs help cluster calls with complex shape. Review your clusters with a streamlined Clustering GUI!
For all your tracking needs! TRex https://t.co/7KG1TZWUhX provides markerless identification of up to 100 animals (flies, fish, locusts, termites, mice, whatever): platform independent (including native M1), very fast, and has a rich user interface. Amazing work @po7ooo!
DeepSqueak V3 is coming soon, check out the test build!
https://t.co/v9bkbrRJUY
𝗩𝟯 𝗠𝗮𝗷𝗼𝗿 𝗨𝗽𝗴𝗿𝗮𝗱𝗲𝘀!
-Brand New YOLO V2 Networks
-Load & Navigate Entire Audio Files
-Hand Box Calls and Train a New Species Detector
-VAE Clustering
Test out the new version of @DeepSqueak_USV!
https://t.co/dobwDe2Ps9
Navigate entire audio files, add new boxes, and create new training sets from scratch. Plus more to come 😉🐬
This is very much in beta, so discuss issues with the community on https://t.co/OdE1SUrBW9