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