Discovering Human Interactions in Videos with Limited Data Labeling

We present a novel approach for discovering human in-teractions in videos. Activity understanding techniques usu-ally require a large number of labeled examples, which are not available in many practical cases. Here, we focus on re-covering semantically meaningful clusters of human-human and human-object interaction in an unsupervised fashion. A new iterative solution is introduced based on Maximum Margin Clustering (MMC), which also accepts user feed-back to refine clusters. This is achieved by formulating the whole process as a unified constrained latent max-margin clustering problem. Extensive experiments have been car-ried out over three challenging datasets, Collective Activity, VIRAT, and UT-interaction. Empirical results demonstrate that the proposed algorithm can efficiently discover perfect semantic clusters of human interactions with only a small amount of labeling effort. 1