Jan van Gemert

Equivariance helps neural networks recognize images consistently despite transformations.

This study explores how smaller receptive fields enhance action recognition in videos.

This study evaluates existing video datasets for long-term action recognition effectiveness.

A look at how TAL models work with limited data and computing power.

SSIG provides a simpler method to compare floor plan similarities effectively.

Using machine learning to predict which video segments viewers replay the most.

A new dataset to improve architectural floor plan generation methods for complex buildings.

Research shows people prefer larger bounding boxes in object detection.

Study reveals potential in reducing event data for better efficiency.