Deep Spatio-temporal Manifold Network for Action Recognition

TL;DR

This paper introduces a deep spatio-temporal manifold network that leverages manifold structures to improve action recognition in videos by constraining feature learning, reducing overfitting, and enhancing performance.

Contribution

It proposes a novel manifold-constrained deep learning framework for action recognition, integrating manifold priors into CNN training via ADMM-BP, which is theoretically justified and empirically effective.

Findings

Significant performance improvements on benchmark datasets.

Effective reduction of intra-class variations and overfitting.

Theoretical proof of manifold projection in the learning process.

Abstract

Visual data such as videos are often sampled from complex manifold. We propose leveraging the manifold structure to constrain the deep action feature learning, thereby minimizing the intra-class variations in the feature space and alleviating the over-fitting problem. Considering that manifold can be transferred, layer by layer, from the data domain to the deep features, the manifold priori is posed from the top layer into the back propagation learning procedure of convolutional neural network (CNN). The resulting algorithm --Spatio-Temporal Manifold Network-- is solved with the efficient Alternating Direction Method of Multipliers and Backward Propagation (ADMM-BP). We theoretically show that STMN recasts the problem as projection over the manifold via an embedding method. The proposed approach is evaluated on two benchmark datasets, showing significant improvements to the baselines.