Dynamic Texture Video Classification Using Extreme Learning Machine

Wang, Liuyang; Liu, Huaping; Sun, Fuchun

doi:10.1007/978-3-319-14066-7_5

Cited by 4 publications

(4 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Here, the global model is replaced by a set of codebook LDS models, each describing a small space-time cuboid, used as local descriptors in a bag-of-words framework. For additional LDS-based approaches, see also Chan and Vascon-celos [4], Mumtaz et al [53], Qiao and Weng [59], Wang et al [75] and Sagel and Kleinsteuber [65].…”

Section: Related Workmentioning

confidence: 99%

Dynamic Texture Recognition Using Time-Causal and Time-Recursive Spatio-Temporal Receptive Fields

Jansson

Lindeberg

2018

J Math Imaging Vis

View full text Add to dashboard Cite

This work presents a first evaluation of using spatio-temporal receptive fields from a recently proposed time-causal spatiotemporal scale-space framework as primitives for video analysis. We propose a new family of video descriptors based on regional statistics of spatio-temporal receptive field responses and evaluate this approach on the problem of dynamic texture recognition. Our approach generalises a previously used method, based on joint histograms of receptive field responses, from the spatial to the spatio-temporal domain and from object recognition to dynamic texture recognition. The time-recursive formulation enables computationally efficient time-causal recognition. The experimental evaluation demonstrates competitive performance compared to state of the art. In particular, it is shown that binary versions of our dynamic texture descriptors achieve improved performance compared to a large range of similar methods using different primitives either handcrafted or learned from data. Further, our qualitative and quantitative investigation into parameter choices and the use of different sets of receptive fields highlights the robustness and flexibility of our approach. Together, these results support the descriptive power of this family of time-causal spatio-temporal receptive fields, validate our approach for dynamic texture recognition and point towards the possibility of designing a range of video analysis methods based on these new time-causal spatio-temporal primitives.

show abstract

Section: Related Workmentioning

confidence: 99%

Dynamic Texture Recognition Using Time-Causal and Time-Recursive Spatio-Temporal Receptive Fields

Jansson

Lindeberg

2018

J Math Imaging Vis

View full text Add to dashboard Cite

show abstract

“…Then, the bag-of-system-trees was further proposed for better efficiency [11]. Extreme learning machine (ELM) was applied to construct the codebook of LDS features while preserving the spatial and temporal characteristics of dynamic textures [12]. A hierarchical expectation maximization algorithm was proposed to cluster dynamic textures using LDS features [13].…”

Section: A Dynamic Texture Recognitionmentioning

confidence: 99%

“…All of these datasets have a relatively large resolution and complex background, which makes VF more robust than AF. However, AF outperforms Methods Beta Gamma DFS [18] 76.9 74.8 MBSIF-TOP [16] 90.7 91.3 ELM [12] 93.7 88.3 ST-TCoF [27] 98. Methods Beta Gamma DFS [18] 76.5 74.5 OTF [47] 75.4 73.5 LBP-TOP [25] 73.4 72.0 OTD [45] 76 VF on the DynTex++ dataset.…”

Section: Feature Evaluationmentioning

confidence: 99%

Manifold Regularized Slow Feature Analysis for Dynamic Texture Recognition

Miao¹,

Xu²,

Xing³

et al. 2017

Preprint

View full text Add to dashboard Cite

Dynamic textures exist in various forms, e.g., fire, smoke, and traffic jams, but recognizing dynamic texture is challenging due to the complex temporal variations. In this paper, we present a novel approach stemmed from slow feature analysis (SFA) for dynamic texture recognition. SFA extracts slowly varying features from fast varying signals. Fortunately, SFA is capable to leach invariant representations from dynamic textures. However, complex temporal variations require highlevel semantic representations to fully achieve temporal slowness, and thus it is impractical to learn a high-level representation from dynamic textures directly by SFA. In order to learn a robust low-level feature to resolve the complexity of dynamic textures, we propose manifold regularized SFA (MR-SFA) by exploring the neighbor relationship of the initial state of each temporal transition and retaining the locality of their variations. Therefore, the learned features are not only slowly varying, but also partly predictable. MR-SFA for dynamic texture recognition is proposed in the following steps: 1) learning feature extraction functions as convolution filters by MR-SFA, 2) extracting local features by convolution and pooling, and 3) employing Fisher vectors to form a video-level representation for classification. Experimental results on dynamic texture and dynamic scene recognition datasets validate the effectiveness of the proposed approach.

show abstract

“…In [ 22 ], a discriminative learning method is developed to formulate the classification problem on Riemannian space by covd, which presents a kernel function and a log-Euclidean distance metric to solve Riemannian-Euclidean transformation. In [ 23 ], a coding strategy is introduced, and the descriptor can be transformed into a new feature; and then, extreme learning machine (ELM) can be used for dynamic texture video classification. However, such a method separately optimizes the reconstruction error of the coding and the classification error of ELM, and the design stage of coding and the classifier are totally independent.…”

Section: Introductionmentioning

confidence: 99%

Scene-Level Geographic Image Classification Based on a Covariance Descriptor Using Supervised Collaborative Kernel Coding

Yang

Liu

Wang

et al. 2016

Sensors

Self Cite

View full text Add to dashboard Cite

Scene-level geographic image classification has been a very challenging problem and has become a research focus in recent years. This paper develops a supervised collaborative kernel coding method based on a covariance descriptor (covd) for scene-level geographic image classification. First, covd is introduced in the feature extraction process and, then, is transformed to a Euclidean feature by a supervised collaborative kernel coding model. Furthermore, we develop an iterative optimization framework to solve this model. Comprehensive evaluations on public high-resolution aerial image dataset and comparisons with state-of-the-art methods show the superiority and effectiveness of our approach.

show abstract

Dynamic Texture Video Classification Using Extreme Learning Machine

Cited by 4 publications

References 20 publications

Dynamic Texture Recognition Using Time-Causal and Time-Recursive Spatio-Temporal Receptive Fields

Dynamic Texture Recognition Using Time-Causal and Time-Recursive Spatio-Temporal Receptive Fields

Manifold Regularized Slow Feature Analysis for Dynamic Texture Recognition

Scene-Level Geographic Image Classification Based on a Covariance Descriptor Using Supervised Collaborative Kernel Coding

Contact Info

Product

Resources

About