Comparative Evaluation of Action Recognition Methods via Riemannian Manifolds, Fisher Vectors and GMMs: Ideal and Challenging Conditions

Abstract

We present a comparative evaluation of various techniques for action recognition while keeping as many variables as possible controlled. We employ two categories of Riemannian manifolds: symmetric positive definite matrices and linear subspaces. For both categories we use their corresponding nearest neighbour classifiers, kernels, and recent kernelised sparse representations. We compare against traditional action recognition techniques based on Gaussian mixture models and Fisher vectors (FVs). We evaluate these action recognition techniques under ideal conditions, as well as their sensitivity in more challenging conditions (variations in scale and translation). Despite recent advancements for handling manifolds, manifold based techniques obtain the lowest performance and their kernel representations are more unstable in the presence of challenging conditions. The FV approach obtains the highest accuracy under ideal conditions. Moreover, FV best deals with moderate scale and translation changes.

An open source C++ implementation of multi-threaded Gaussian mixture models, k-means and expectation maximisation

“…(3) can be executed independently for each sample, the summations in Eqns. (4) and (6) can be split into separate sets of summations, where the summation in each set can be executed independently and in parallel with other sets. To allow similar splitting of the summation for calculating covariance matrices, Eqn.…”

“…Modelling multivariate data through a convex mixture of Gaussians, also known as a Gaussian mixture model (GMM), has many uses in fields such as signal processing, econometrics, pattern recognition, machine learning and computer vision. Examples of applications include multi-stage feature extraction for action recognition [4], modelling of intermediate features derived from deep convolutional neural networks [11,12,16], classification of human epithelial cell images [32], implicit sparse coding for face recognition [33], speech-based identity verification [28], and probabilistic foreground estimation for surveillance systems [26]. GMMs are also commonly used as the emission distribution for hidden Markov models [2].…”

Fuzzy Empowered Cognitive Spatial Relation Identification and Semantic Action Recognition

“…They developed tools with smoothing, denoising, temporal registration and extraction of action in a time domain. Johanan et al [3] compare the Gaussian mixture model based action recognition algorithm with Fisher vectors model with symmetric positive definite matrices and linear subspaces. In their evaluation Fisher vector model obtains higher accuracy rate for scale invariant and ideal condition.…”