Fernando De la Torre

Data

Quality of Life Grand Challenge: Kitchen Capture

Multimodal database of subjects performing the tasks involved in cooking captured with several sensors (audio, video, motion capture, accelerometers / gyroscopes).

Multimodal Action Database (MAD)

The MAD database contains multimodal recordings (video, depth, skeleton) of 35 activities of 20 subjects with annotations for the start and the end of each action. The MAD database is particularly tailored to address activity detection (not classification) in video.

Code

Robust Principal Component Analysis (17MB)

Feature Selection for SVM

Joint Localization and Classification

Max-Margin Early Event Detectors

Aligned Cluster Analysis (ACA)

Canonical Time Warping (CTW)

Generalized Time Warping (GTW)

Factorized Graph Matching (FGM)

Temporal Commonality Discovery

Intraface