@MIT researchers have developed an innovative AI technique to identify specific actions in lengthy videos! This method from @MITIBMLab trains models to understand unannotated films by analyzing minute details and overall context. It pinpoints the exact times and locations of
MULTIMODAL AI
-
Advancements in generative AI video synthesis capabilities
By
–
I think that previously it was doing an okay job with animating images but now it has become much better in video gen from text. This is a big difference.
-
Pika Labs Releases New Text-to-Video AI Model
By
–
BREAKING: Pika Labs released a new text-to-video model. All generations have a very very good quality from the first prompt 🔥
— 🚨 AI News | TestingCatalog (@testingcatalog) 11 juin 2024
"Cyberpunk hacker" example I've just made. All vids are 3 seconds long but can be enhanced further. https://t.co/JdFyEs7kOP pic.twitter.com/kSdHyLR0jOBREAKING: Pika Labs released a new text-to-video model. All generations have a very very good quality from the first prompt "Cyberpunk hacker" example I've just made. All vids are 3 seconds long but can be enhanced further.
-
Google releases new generative AI image alphabet experiment
By
–
ICYMI: Google released a new Image Gen experiment which allows you to generate a themed alphabet and then use it to type any text.
— 🚨 AI News | TestingCatalog (@testingcatalog) 11 juin 2024
This would be huge if imagen 2 would be that good at writing text 🔥 https://t.co/bG5w7sI6EM pic.twitter.com/Nj681TznkMICYMI: Google released a new Image Gen experiment which allows you to generate a themed alphabet and then use it to type any text. This would be huge if imagen 2 would be that good at writing text
-
Image-to-Video Model Update Represents Major AI Advancement
By
–
This image-to-video model update is wild. pic.twitter.com/L7117cMnNO
— Pika (@pika_labs) 11 juin 2024This image-to-video model update is wild.
-
Google MUM Multilingual AI Model from IO 2018
By
–
It just remember me the google io 18, when they talk about Mum and multilinguage…
-
Diffusion Models Reconstruct Fragmented 3D Objects
By
–
Qué chulo este trabajo que usa modelos de difusión para reconstruir objetos tridimensionales fragmentados y resolver el puzzle de armarlos.pic.twitter.com/KlvOxvlbqp
— Carlos Santana (@DotCSV) 11 juin 2024Qué chulo este trabajo que usa modelos de difusión para reconstruir objetos tridimensionales fragmentados y resolver el puzzle de armarlos.
-
Affective Behavior Analysis in the Wild Conference 2026
By
–
…
https://affective-be
havior-analysis-in-the-wild.github.io/6th/ -
CVPR Keynote on Affective Behavior Analysis Workshop
By
–
Excited to attend @CVPR next week and give a keynote at Affective Behavior Analysis in the Wild workshop. It would be great to catch up if you're around!
-
User commentary on AI video generation competitiveness
By
–
C’est tellement étonnant Sora doit être en PLS