Bag-of-words with aggregated temporal pair-wise word co-occurrence for human action recognition

Agustí, Pau; Traver, Vicente; Pla, Filiberto

doi:10.1016/j.patrec.2014.07.014

Cited by 10 publications

(5 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Bag of words (BoW) [40] is most likely used as a feature representation scheme for the still images and the videos in HAR. The bag of visual words, also known as the BoW, is a symbolic scheme used to symbolize the documents for retrieval purposes.…”

Section: Bowmentioning

confidence: 99%

See 1 more Smart Citation

Ensemble Based Feature Extraction and Deep Learning Classification Model with Depth Vision

Sinha,

Kumar,

Ghosh

2023

cai

View full text Add to dashboard Cite

It remains a challenging task to identify human activities from a video sequence or still image due to factors such as backdrop clutter, fractional occlusion, and changes in scale, point of view, appearance, and lighting. Different appliances, as well as video surveillance systems, human-computer interfaces, and robots used to study human behavior, require different activity classification systems. A four-stage framework for recognizing human activities is proposed in the paper. As part of the initial stages of pre-processing, video-to-frame conversion and adaptive histogram equalization (AHE) are performed. Additionally, watershed segmentation is performed and, from the segmented images, local texton XOR patterns (LTXOR), motion boundary scale-invariant feature transforms (MoBSIFT) and bag of visual words (BoW) based features are extracted. The Bidirectional gated recurrent unit (Bi-GRU) and the Bidirectional long short-term memory (Bi-LSTM) classifiers are used to detect human activity. In addition, the combined decisions of the Bi-GRU and Bi-LSTM classifiers are further fused, and their accuracy levels are determined. With this Dempster-Shafer theory (DST)

show abstract

Section: Bowmentioning

confidence: 99%

“…The Bi-GRU [40] uses exceptional gates (ut), known as reset and update gates for the declining gradient dispersion with smaller loss. The ut substitute input and forget gate of the LSTM, which depict the preservation degree of the preceding data.…”

Section: Bi-gru Variant Of Rnn Based Recognitionmentioning

confidence: 99%

Ensemble Based Feature Extraction and Deep Learning Classification Model with Depth Vision

Sinha,

Kumar,

Ghosh

2023

cai

View full text Add to dashboard Cite

show abstract

“…In addition, Equation (6) shows an equation for calculating the Fmeasure using accuracy and recall. The performance evaluation uses the proposed miningbased mutual information (MbMI), existing mining-based word frequency (MbWF) [37,38], word concurrence frequency (WCoF) [39,40] in the document to find the relationship between words. It performs performance evaluation while repeatedly changing minimum support.…”

Section: B Performance Evaluationmentioning

confidence: 99%

Multi-Level Health Knowledge Mining Process in P2P Edge Network

Baek

Chung

2021

IEEE Access

View full text Add to dashboard Cite

Chronic diseases are increasing due to westernized eating habits and everyday life changes, and healthcare and disease prevention should be managed based on constant interest. Users, who are not health professionals, have difficulty in obtaining accurate information related to healthcare due to noise problems such as subjective opinions, distorted information, and exaggerated information. There is a need for a method that enables users to obtain meaningful information for healthcare and disease prevention in real-time among the vast amounts of data collected through search. In this study, we propose a multi-level health knowledge mining process in a P2P edge network. The proposed method suggests a P2P edge network to solve the overload problem of P2P networking, the noise problem, and the security problem of cloud computing and mines the health knowledge through the mutual information according to the association rules. In addition, the results of health knowledge mining are visualized to propose a method by which users can easily receive relevant health information. As a result of the performance evaluation, the F-measure using recall and precision is 83%, 79%, 75%, 74%, and 73% of the support ratings of 10%, 20%, 30%, 40%, and 50% Was derived. Accordingly, it is possible to process and analyze healthcare-related information in realtime through a multi-level based health knowledge tree based on the association of data collected by P2P edge computing. In addition, by visualizing meaningful information to the user through the embedding network structure, it provides personalized information for intuitive understanding.

show abstract

“…Since the effectiveness of the hand-crafted features directly influences the overall recognition performance, a lot of well-designed features were proposed, whose instances include HoG [11], histogram of optical flows (HoF) [13], 3D Harris corners [14], 3D scale invariant feature transform (3D SIFT) [15], and so on. These features are generally extracted from spatiotemporal local regions in an input video and integrated into a video-wise feature vector with bags of visual word (BoVW) strategy [16].…”

Section: A Supervised Activity Video Analysismentioning

confidence: 99%

Semi-Supervised Temporal Segmentation of Manufacturing Work Video by Automatically Building a Hierarchical Tree of Category Labels

Nakamura¹,

Nitta²,

Babaguchi³

et al. 2021

IEEE Access

View full text Add to dashboard Cite

Nowadays, many industrial companies visually record workers' activities for the purposes of streamlining their work processes. However, since untrimmed raw videos are hard to use, it is desired to automatically divide the videos into segments and recognize which kind of operation is performed on each segment. This task is called temporal video segmentation. We propose a method for achieving it, particularly targeting videos of manufacturing work with a specialized vehicle such as a hydraulic excavator. To make the performance of temporal video segmentation high, it is quite essential to extract good visual features from input videos. This can be hardly achieved by unsupervised methods, whereas supervised methods have another drawback that collecting a sufficient amount of training data is difficult due to its labor-intensiveness. To overcome these drawbacks, the proposed method employs a semi-supervised approach. We assume that a set of weakly-labeled videos whose frames only sparsely have a category label are given as input, where the labeled frames are used as training data to train a desirable feature extractor. Under this assumption, the proposed method first divides the input videos into short segments called primitive segments having the fixed length and then clusters them using visual features extracted by the above feature extractor. To achieve higher performance, we also use a hierarchical tree of the category labels and recursively perform the above process at each branch in the tree, where the tree is automatically built by the proposed method. In our experiments, we achieved a segmentation performance of 0.947 on the F-measure, even when only 1.25% of all the frames in the input videos are labeled.INDEX TERMS temporal video segmentation, semi-supervised clustering, hierarchical tree of category labels, manufacturing work video, deep metric learning

show abstract

Bag-of-words with aggregated temporal pair-wise word co-occurrence for human action recognition

Cited by 10 publications

References 40 publications

Ensemble Based Feature Extraction and Deep Learning Classification Model with Depth Vision

Ensemble Based Feature Extraction and Deep Learning Classification Model with Depth Vision

Multi-Level Health Knowledge Mining Process in P2P Edge Network

Semi-Supervised Temporal Segmentation of Manufacturing Work Video by Automatically Building a Hierarchical Tree of Category Labels

Contact Info

Product

Resources

About