Learning Hierarchical Invariant Sptio-temporal Features for Action Recognition with Independent Subspace Analysis. This package is aimed at providing read-to-use video features the stacked convolutional ISA model described in the paper, and functions