Leonid Sigal

A new approach for improving machine learning systems in image recognition tasks.

A new method improves few-shot segmentation using visual prompts and attention mechanisms.

New method improves the creation of lifelike 3D avatars from video footage.

A new model improves how machines read charts, even without labels.

Introducing Long Video Masked Autoencoders for better video understanding.

Learn how new models are making video generation faster and better.

Researchers develop benchmarks for vision-language models to reason about unexpected events in videos.

Learn how the HIST framework improves image and text understanding.

A new method improves adversarial image creation in medical imaging.

Learn how models adapt to new data without original labels using innovative techniques.

A framework that simplifies visual task solutions for everyone.