This week in AI & Machine Learning: A pivotal year for computer vision, SEER-10B, machine learning on tractors, and more.
Artificial Intelligence News:
Computer Vision and Digital Transformation: Plainsight’s CEO Joins the Alldus ‘AI In Action’ Podcast
Carlos Anchia, Co-Founder and CEO of Plainsight, joined the ‘AI in action’ podcast series to discuss the circumstances that promise to make 2022 a pivotal year for computer vision. Listen to the full podcast here.
“The models are more accurate than they were five years ago, the edge infrastructure is more powerful and more cost effective than it ever was, we have this environment where everything is imagery . . . computer vision in 2022 is going to explode because of these enabling factors.”
– Carlos Anchia
SEER 10B: Better, fairer computer vision through self-supervised learning on diverse datasets
Last year Meta announced SEER, a computer vision model that outperformed most state-of-the-art systems. This week, SEER-10B was announced, promising groundbreaking results with a whopping 10 billion dense parameters, making it potentially the largest neural network for computer vision. Read more on the official meta AI blog.
We’re pleased to announce new advances in SEER, Meta AI’s groundbreaking self-supervised #computervision model. SEER is now not only much more powerful, it also produces fairer, more robust computer vision models. Learn more: https://t.co/OferzHr9Ic pic.twitter.com/xEDOY34Qls
— Meta AI (@MetaAI) February 28, 2022
Deploying Machine Learning Models at Scale to Tractors
A huge part of the machine learning lifecycle is managing model deployments. Learn how Alon from Greeneye and Moses from ClearML deploy models to tractors! Listen to the podcast on changelog.
Mark Zuckerberg: Meta, AI, Facebook, Instagram, and the Metaverse
Mark Zuckerberg joined Lex Fridman to discuss his thoughts on technology connecting people in a rare free-form interview appearance. Now we’ve all seen that Zuckerberg can pass a captcha test!
Co-training Transformer with Videos and Images Improves Action Recognition
Using computer vision for action recognition has become a widely researched topic. The latest state-of-the-art results from Google come from combining video and image data to co-train a transformer model. Read the full story on the official Google blog.