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

@inproceedings{Carvajal2016ComparativeEO,
  title={Comparative Evaluation of Action Recognition Methods via Riemannian Manifolds, Fisher Vectors and GMMs: Ideal and Challenging Conditions},
  author={Johanna Carvajal and Arnold Wiliem and Chris McCool and Brian C. Lovell and Conrad Sanderson},
  booktitle={PAKDD Workshops},
  year={2016}
}
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… 
Human Action Recognition using Ensemble of Shape, Texture and Motion features
TLDR
A new feature descriptor is introduced which is named ‘Spatio Temporal Shape-Texture-Motion’ (STSTM) descriptor which uses hybrid approach by combining local and global features.
A Novel Human Action Recognition and Behaviour Analysis Technique using SWFHOG
TLDR
The proposed SWFHOG method shows promising results as compared to earlier methods, and is tested against Camera view angle change and imperfect actions using Weizmann robustness testing datasets.
Action analysis and video summarisation to efficiently manage and interpret video data
TLDR
A novel approach to video summarisation that makes use of a Bag-of-visualTextures approach which is computationally efficient and effective and can be used for short-term and long-term videos.
An open source C++ implementation of multi-threaded Gaussian mixture models, k-means and expectation maximisation
  • C. Sanderson, Ryan R. Curtin
  • Computer Science
    2017 11th International Conference on Signal Processing and Communication Systems (ICSPCS)
  • 2017
TLDR
It is demonstrated that the multi-threaded implementation of Gaussian mixture models in the C++ language achieves a speedup of an order of magnitude on a recent 16 core machine, and that it can achieve higher modelling accuracy than a previously well-established publically accessible implementation.
Fuzzy Empowered Cognitive Spatial Relation Identification and Semantic Action Recognition
TLDR
This paper proposed a fuzzy-based action recognition system from a basketball sports image which would automatically and semantically label the action in the videos through machine understandability concept.
Indonesian Journal of Electrical Engineering and Computer Science
TLDR
SDMM algorithm detects and mitigates prolonged Slowloris attack in typical networks as well as in slow networks with low bandwidth and high delay in 240-280s with an accuracy of 100% and 98% respectively.
An Improved Model for Human Activity Recognition by Integrated feature Approach and Optimized SVM
TLDR
An integrated features approach using Histogram of Gradient (HOG) local feature descriptor and Principal Component Analysis (PCA) as a global feature is proposed in this paper to automatically recognize human activity from a video sequence.

References

SHOWING 1-10 OF 57 REFERENCES
Kernel analysis on Grassmann manifolds for action recognition
Kernel analysis over Riemannian manifolds for visual recognition of actions, pedestrians and textures
TLDR
Experiments on several visual classification tasks show that the proposed embedding into the Reproducing Kernel Hilbert Space by introducing a Riemannian pseudo kernel obtains considerable improvements in discrimination accuracy.
Action Recognition in Video by Sparse Representation on Covariance Manifolds of Silhouette Tunnels
TLDR
A novel framework for action recognition in video using empirical covariance matrices of bags of low-dimensional feature vectors is developed, which attains a segment-level classification rate of 96.74% for the Weizmann dataset and 96.15%" for the UT-Tower dataset.
Action Recognition From Video Using Feature Covariance Matrices
TLDR
A general framework for fast and accurate recognition of actions in video using empirical covariance matrices of features, which is conceptually simple and has low storage and computational requirements making it attractive for real-time implementation.
Tangent Bundles on Special Manifolds for Action Recognition
  • Y. Lui
  • Computer Science
    IEEE Transactions on Circuits and Systems for Video Technology
  • 2012
TLDR
The empirical results reveal that the proposed paradigm frames the classification scheme as a nearest neighbor based on the intrinsic distance, prior training is unnecessary and the method is highly competitive with the current state-of-the-art methods, robust to small alignment errors, and yet simpler.
Human gesture recognition on product manifolds
  • Y. Lui
  • Computer Science
    J. Mach. Learn. Res.
  • 2012
TLDR
The aim of this paper is to demonstrate the importance of the intrinsic geometry of tensor space which yields a very discriminating structure for action recognition and it is shown that the proposed method performs well on the standard benchmark data sets, but also it generalizeswell on the one-shot-learning gesture challenge.
Tangent bundle for human action recognition
TLDR
This work introduces a new method for automated human action recognition that represents videos as a tangent bundle on a Grassmann manifold and achieves recognition rates of 91% on the Cambridge gesture dataset, 88%" on the UCF sport dataset, and 97% onThe KTH human action dataset.
Optimizing over Radial Kernels on Compact Manifolds
TLDR
It is shown that positive definite radial kernels on the unit n-sphere, the Grassmann manifold and Kendall's shape manifold can be expressed in a simple form whose parameters can be automatically optimized within a support vector machine framework.
Covariance discriminative learning: A natural and efficient approach to image set classification
TLDR
A novel discriminative learning approach to image set classification by modeling the image set with its natural second-order statistic, i.e. covariance matrix, which shows the superiority of this method over state-of-the-art ones in both accuracy and efficiency, but also its stability to two real challenges: noisy set data and varying set size.
Random projections on manifolds of Symmetric Positive Definite matrices for image classification
TLDR
Experiments on face recognition, person re-identification and texture classification show that the proposed approach outperforms several recent methods, such as Tensor Sparse Coding, Histogram Plus Epitome, Riemannian Locality Preserving Projection and Relational Divergence Classification.
...
...