Alexei A. Efros

CrossMAE improves image reconstruction efficiency without relying on self-attention.

New methods simplify creating distinct 3D scenes from text descriptions.

Examining the role of neurons in CLIP models and their interactions.

A new perspective on improving image creation through score distillation sampling.

Customizing generative models to reflect unique identities through weight space.

This study presents a new method for identifying key training images in AI-generated visuals.

A fresh method transforms how we analyze large collections of images.

Study compares human and AI abilities in recognizing 3D shapes from different views.