LAMP

The dataset used in the paper is not explicitly described, but it is mentioned that the authors used a dataset that comprises 24 videos with diverse motion types and contexts.

BibTex: