Alexandre Alahi

A new dataset and framework for tracking soccer players using single-camera video footage.

A novel approach to improve accuracy in identifying occluded individuals.

This method improves 3D pose accuracy from 2D images using a transformer network.

SoccerNet 2024 challenges drive innovation in video understanding for soccer.

A look at how machines learn to predict human actions.

Helvipad provides depth information from 360-degree images, aiding machine learning.

GEM transforms video prediction and object interaction with innovative technology.

MotionMap offers a new way to predict human movement accurately.