Suha Kwak

This study enhances video action detection by focusing on context and classification.

MATR enhances action detection in unedited video streams through memory-augmented technology.

R-Adapter enhances efficiency and robustness in model fine-tuning for image and text tasks.

A new method aims to reduce bias in machine learning models for better fairness.

A new method enhances finding individuals in images using text cues.

A look at how machines learn to recognize objects without labels.

A new model combines action segmentation and anticipation for smarter interactions.