Abhinav Shrivastava

LiFT improves Vision Transformers for detailed image tasks efficiently.

A method for assessing artistic style in generated images.

A new approach to enhance image-text matching using dual-encoder models and scene graphs.

V-VIPE enhances 3D pose estimation from 2D images, overcoming angle challenges.

InVi allows easy addition of objects in videos using existing tools.

WayEx helps robots learn tasks efficiently with fewer demonstrations.

A new approach merges motion and appearance for effective action recognition with limited data.

A new method enables machines to accurately model moving and changing shapes.

New methods speed up video encoding and decoding.

VeriGraph enhances robot cleaning efficiency through advanced scene graph technology.

A new framework makes streaming dynamic 3D videos faster and more efficient.

Learn how video decomposition transforms editing for creators and filmmakers.

New methods improve video predictions using less data.