Human action recognition-based video summarization for RGB-D personal sports video

Tejero-de-Pablos, Antonio; Nakashima, Yuta; Sato, Tomokazu; Yokoya, Naokazu

doi:10.1109/icme.2016.7552938

Cited by 14 publications

(26 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The best combination among the evaluated features corresponds to a combination of 3D body joint-based features and CNN-ISA features [14]). In contrast to the previous work [12], the results indicate that it is not necessary to explicitly recognize players' actions in order to determine highlights. Alternatively, deep neural networks are leveraged to extract a feature representation of players' actions and to model their temporal dependency.…”

Section: B Resultscontrasting

confidence: 93%

“…1) Objective evaluation by segment f-score: The ability of the proposed method to extract highlights was evaluated in terms of the f-score. In the proposed method, a one-second period of video is as follows: [4] 0.27 0.65 GMM-HMM [12] 0.44 0.79…”

Section: B Resultsmentioning

confidence: 99%

“…Table II shows the f-scores for the summaries generated with the labels of both E and NE groups. In addition to the features described in Section IV-A, it includes the results of using the features from our previous work in UGSV summarization [12]. The features were obtained by feeding the 3D body joint representation of players' actions to the action recognition method in [59] and considering the action classification results.…”

Section: B Resultsmentioning

confidence: 99%

“…Hence, in this work, we hypothesize that features extracted from players' actions allow summarizing UGSV. The method in [12] is extended by employing a two-stream deep neural network [19], [20]. Our new method considers two different types of inputs, namely RGB frames of video and body joint positions, and each of them are transformed through two separate neural networks (i.e., streams).…”

Section: Ugsv Summarization Using Action Featuresmentioning

confidence: 99%

See 3 more Smart Citations

Summarization of User-Generated Sports Video by Using Deep Action Recognition Features

Tejero-de-Pablos

Nakashima

Sato

et al. 2018

IEEE Trans. Multimedia

Self Cite

View full text Add to dashboard Cite

Section: B Resultscontrasting

confidence: 93%

Section: B Resultsmentioning

confidence: 99%

Section: B Resultsmentioning

confidence: 99%

Section: Ugsv Summarization Using Action Featuresmentioning

confidence: 99%

See 2 more Smart Citations

Summarization of User-Generated Sports Video by Using Deep Action Recognition Features

Tejero-de-Pablos

Nakashima

Sato

et al. 2018

IEEE Trans. Multimedia

Self Cite

View full text Add to dashboard Cite

“…Multimodal data has been studied for a variety of applications to analyze human behaviors, including person detection and identification [9,10], human action recognition [11,12], face recognition [13,14], as well as sentiment analysis.…”

Section: Related Workmentioning

confidence: 99%

Select-additive learning: Improving generalization in multimodal sentiment analysis

Wang

Meghawat

Morency

et al. 2017

2017 IEEE International Conference on Multimedia and Expo (ICME)

170

View full text Add to dashboard Cite

Multimodal sentiment analysis is drawing an increasing amount of attention these days. It enables mining of opinions in video reviews which are now available aplenty on online platforms. However, multimodal sentiment analysis has only a few high-quality data sets annotated for training machine learning algorithms. These limited resources restrict the generalizability of models, where, for example, the unique characteristics of a few speakers (e.g., wearing glasses) may become a confounding factor for the sentiment classification task. In this paper, we propose a Select-Additive Learning (SAL) procedure that improves the generalizability of trained neural networks for multimodal sentiment analysis. In our experiments, we show that our SAL approach improves prediction accuracy significantly in all three modalities (verbal, acoustic, visual), as well as in their fusion. Our results show that SAL, even when trained on one dataset, achieves good generalization across two new test datasets.

show abstract

The Rise of Machine Learning in Polymer Discovery

Yan

2023

Advanced Intelligent Systems

View full text Add to dashboard Cite

In the recent decades, with rapid development in computing power and algorithms, machine learning (ML) has exhibited its enormous potential in new polymer discovery. Herein, the history of ML is described and the basic process of ML accelerated polymer discovery is summarized. Next, the four steps in this process are reviewed, that is, dataset selection, fingerprinting, ML framework, and new polymer generation. Finally, a couple of main challenges for ML accelerated polymer discovery is presented and the outlooks in this field are prospected. It is expected that this review can service as a useful tool for the people who just step into this field and deepen the understanding for the people who are already in this field.

show abstract

Human action recognition-based video summarization for RGB-D personal sports video

Cited by 14 publications

References 23 publications

Summarization of User-Generated Sports Video by Using Deep Action Recognition Features

Summarization of User-Generated Sports Video by Using Deep Action Recognition Features

Select-additive learning: Improving generalization in multimodal sentiment analysis

The Rise of Machine Learning in Polymer Discovery

Contact Info

Product

Resources

About