Dimitris Samaras

A model predicting viewer focus on graphic design documents.

New method improves image segmentation accuracy using patch-based classification.

A novel method enhances semi-supervised segmentation by focusing on reliable pseudo-labels.

A model designed for creating large, high-quality images efficiently.

Introducing an efficient way to evaluate the quality of generated samples using latent density scores.

A new model predicts where people look based on spoken commands.

A new method blends audio and facial expressions for realistic video generation.

A new system models emotional intensity in animated characters for enhanced realism.

A system helps computers match images with complex descriptions effectively.

Learn how ZoomLDM improves image quality across different zoom levels.

Revolutionizing machine learning with self-generated image variations.