Gesture Modeling by Hanklet-Based Hidden Markov Model

Presti, Liliana Lo; Cascia, Marco La; Sclaroff, Stan; Camps, Octavia

doi:10.1007/978-3-319-16811-1_35

Cited by 23 publications

(38 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the NTU RGB+D experiments, φ φ φ P improves (by margin) Fisher vectors [49], Lie group representation [7] as [22] 92.2 64.0 T-approx [20] 92.8 62.0 φ φ φ kron−e (proposed) 92.3 65.0 φ φ φ kron−π (proposed) 95.6 66.5 t-COV-pyramid [30] 89.2 -Bregman-div [3] 89.9 58.2 Ker-RP-RBF [2] 92.3 66.2 J-DI E -CNN [14] 93.1 − Ker-COV [4] 95.0 rnd-logHS [17] 97. [22] 88.2 88.6 84.0 T-approx [20] 89.6 88.9 84.0 φ φ φ kron−e (proposed) 89.5 89.6 84.4 φ φ φ kron−π (proposed) 89.9 89.9 84.0 t-COV-pyramid [30] 74.0 91.5 -H-HMM [32] 89.0 -86.8 rnd-logHS [17] 91.5 88.5 89.7 QMC-logHS [17] 90 Classification accuracies [%] for 3D action recognition. For each table, the top part present the performance achieved by φ φ φ kron−π and φ φ φ kron−e against other alternative approximating schemes [19], [20], [21], [22], [29]: within this class of methods, the best accuracy is highlighted in bold.…”

Section: Discussionmentioning

confidence: 99%

Scalable and compact 3D action recognition with approximated RBF kernel machines

Cavazza

Morerio

Murino

2019

Pattern Recognition

View full text Add to dashboard Cite

Despite the recent deep learning (DL) revolution, kernel machines still remain powerful methods for action recognition. DL has brought the use of large datasets and this is typically a problem for kernel approaches, which are not scaling up efficiently due to kernel Gram matrices. Nevertheless, kernel methods are still attractive and more generally applicable since they can equally manage different sizes of the datasets, also in cases where DL techniques show some limitations. This work investigates these issues by proposing an explicit approximated representation that, together with a linear model, is an equivalent, yet scalable, implementation of a kernel machine. Our approximation is directly inspired by the exact feature map that is induced by an RBF Gaussian kernel but, unlike the latter, it is finite dimensional and very compact. We justify the soundness of our idea with a theoretical analysis which proves the unbiasedness of the approximation, and provides a vanishing bound for its variance, which is shown to decrease much rapidly than in alternative methods in the literature. In a broad experimental validation, we assess the superiority of our approximation in terms of 1) ease and speed of training, 2) compactness of the model, and 3) improvements with respect to the state-of-the-art performance.

show abstract

Section: Discussionmentioning

confidence: 99%

Scalable and compact 3D action recognition with approximated RBF kernel machines

Cavazza

Morerio

Murino

2019

Pattern Recognition

View full text Add to dashboard Cite

show abstract

“…This is a challenging dataset because many of the actions are highly similar to each other. Comparative analysis We benchmarked the proposed (Log-COV-Net) against the Hankel-based approaches [22,43], used in tandem with either a Hidden Markov Model (HMM) or a Riemannian-nearest neighbors classifier with prototypes (Hankel-NN-proto). Also, we compared against the tensor representation provided by [18] in using Sequence and Dynamics Compatibility Kernels (SCK+DCK).…”

Section: Msr-action 3dmentioning

confidence: 99%

“…Hankel-HMM [22] 89.0% SCK + DCK [18] 94.0% Hankel-NN-proto [43] 94.7% graph-joint-LSTM [21] 94.8% Kernelized-COV [3] 96.8% Ker-RP-RBF [37] 96.9% Log-COV-Net (proposed) 97.4% Table 1. Evaluation on MSR-Action-3D using the protocol of [20].…”

Section: Msr-action 3dmentioning

confidence: 99%

See 1 more Smart Citation

When Kernel Methods Meet Feature Learning: Log-Covariance Network for Action Recognition From Skeletal Data

Cavazza

Morerio

Murino

2017

2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

View full text Add to dashboard Cite

Human action recognition from skeletal data is a hot research topic and important in many open domain applications of computer vision, thanks to recently introduced 3D sensors. In the literature, naive methods simply transfer off-the-shelf techniques from video to the skeletal representation. However, the current state-of-the-art is contended between to different paradigms: kernel-based methods and feature learning with (recurrent) neural networks. Both approaches show strong performances, yet they exhibit heavy, but complementary, drawbacks. Motivated by this fact, our work aims at combining together the best of the two paradigms, by proposing an approach where a shallow network is fed with a covariance representation. Our intuition is that, as long as the dynamics is effectively modeled, there is no need for the classification network to be deep nor recurrent in order to score favorably. We validate this hypothesis in a broad experimental analysis over 6 publicly available datasets.

show abstract

“…Presti et al [22] used linear time invariant (LTI) systems to model subsets of activity demonstrations within fixed size sliding temporal windows, utilizing trained HMMs on the global space of LTIs for classification. A sliding window technique was also used by Gori et al [9], who applied learned 1-D Gaussian filters through time over feature patches extracted from interactions between joint pairs, allowing for quick and robust identification of individual activities and multi-party interactions.…”

Section: Related Workmentioning

confidence: 99%

Interpretable models for fast activity recognition and anomaly explanation during collaborative robotics tasks

Hayes

Shah

2017

2017 IEEE International Conference on Robotics and Automation (ICRA)

View full text Add to dashboard Cite

Abstract-In this paper, we present Rapid Activity Prediction Through Object-oriented Regression (RAPTOR), a scalable method for performing rapid, real-time activity recognition and prediction that achieves state-of-the-art classification accuracy on both a generic human activity dataset and two domainspecific collaborative robotics manufacturing datasets. Our approach is designed to be human-interpretable: able to provide explanations for its reasoning such that non-experts can better understand and improve its activity models. We incorporate methods to increase RAPTOR's resilience against confusion due to temporal variations, as well as against learning false correlations between features. We report full and partial trajectory classification results across three datasets and conclude by demonstrating our model's ability to provide interpretable explanations of its reasoning using outlier detection techniques.

show abstract

Gesture Modeling by Hanklet-Based Hidden Markov Model

Cited by 23 publications

References 35 publications

Scalable and compact 3D action recognition with approximated RBF kernel machines

Scalable and compact 3D action recognition with approximated RBF kernel machines

When Kernel Methods Meet Feature Learning: Log-Covariance Network for Action Recognition From Skeletal Data

Interpretable models for fast activity recognition and anomaly explanation during collaborative robotics tasks

Contact Info

Product

Resources

About