Stefano Soatto

Research reveals ways to improve text-to-image systems efficiently and cost-effectively.

Combining images and text improves accuracy in 3D depth estimation.

New methods improve transforming text into accurate 3D models.

Edit 3D images with precision using various input methods for local changes.

A new method reduces blur in photos caused by atmospheric turbulence.

A new method improves confidence scoring in language models using stable explanations.

A fresh method for creating images from text using specialized models.

Introducing a hybrid memory approach for improved AI performance.

New methods enhance video-language models' understanding of actions, objects, and relationships.

Discover how AI creates stunning visuals from simple text prompts.

Learn how AI is improving memory management for better interactions.