Sania Zahan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Sania Zahan has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Ajmal Mian
The University of Western Australia
2 shared publications - Ghulam Mubashar Hassan
Computer Science and Software Engineering, The University of Western Australia, Perth, Australia
2 shared publications Khademul Islam Molla
Department of Computer Science and Engineering, University of Rajshahi, Rajshahi, Bangladesh
2 shared publicationsAshaduzzaman Milu
Department of Computer Science and Engineering, Varendra University, Rajshahi, Bangladesh
1 shared publicationRabiul Islam
1 shared publicationSabina Yasmin
Department of Computer Science and Engineering, Varendra University, Rajshahi, Bangladesh
1 shared publication
Latest Publications
- 16 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-11
Learning Sparse Temporal Video Mapping for Action Quality Assessment in Floor Gymnastics
Automated athlete performance measurement from sports videos can significantly enhance sports evaluation. It requires modeling extended sequences, as the intricate spatio-temporal progression significantly influences overall performance. However, the lack of comprehensive datasets with long-duration samples has hindered researchers from focusing on temporal aspects, leading them to primarily concentrate on spatial structures for assessing short-duration videos. Consequently, in-depth analysis of longer videos has received limited attention. This study aims to explore long-term videos and analyze local discriminative spatial dependencies and global semantics for sports action quality assessment (AQA). A new dataset, coined AGF-Olympics, is presented in this paper incorporating artistic gymnastic floor routines. It features highly challenging scenarios with extensive variations in background, viewpoint, and scale, with a duration of up to 2 min. Finally, a discriminative non-local attention (DNLA) is introduced for score regression that effectively maps dense feature space to a sparse representation by disentangling complex associations in long-duration sports videos. DNLA encodes crucial features by analyzing cross-space–time correlations and filtering out features with lower significance. Thus, it ensures that the model prioritizes significant joints in the spatial domain and frames in the temporal domain. Experimental results demonstrate that the proposed method achieves superior performances and provides a benchmark for the AGF-Olympics dataset. Overall, the proposed method achieves a 7% higher regression rate with a 65.14% reduction in FLOPS and 52.82% faster inference time compared to the current state-of-the-art method.
- 16 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-11
Learning Sparse Temporal Video Mapping for Action Quality Assessment in Floor Gymnastics
Automated athlete performance measurement from sports videos can significantly enhance sports evaluation. It requires modeling extended sequences, as the intricate spatio-temporal progression significantly influences overall performance. However, the lack of comprehensive datasets with long-duration samples has hindered researchers from focusing on temporal aspects, leading them to primarily concentrate on spatial structures for assessing short-duration videos. Consequently, in-depth analysis of longer videos has received limited attention. This study aims to explore long-term videos and analyze local discriminative spatial dependencies and global semantics for sports action quality assessment (AQA). A new dataset, coined AGF-Olympics, is presented in this paper incorporating artistic gymnastic floor routines. It features highly challenging scenarios with extensive variations in background, viewpoint, and scale, with a duration of up to 2 min. Finally, a discriminative non-local attention (DNLA) is introduced for score regression that effectively maps dense feature space to a sparse representation by disentangling complex associations in long-duration sports videos. DNLA encodes crucial features by analyzing cross-space–time correlations and filtering out features with lower significance. Thus, it ensures that the model prioritizes significant joints in the spatial domain and frames in the temporal domain. Experimental results demonstrate that the proposed method achieves superior performances and provides a benchmark for the AGF-Olympics dataset. Overall, the proposed method achieves a 7% higher regression rate with a 65.14% reduction in FLOPS and 52.82% faster inference time compared to the current state-of-the-art method.
- 10 November 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 19 (8) , 8713-8721
SDFA: Structure-Aware Discriminative Feature Aggregation for Efficient Human Fall Detection in Video
Older people are susceptible to fall due to instability in posture and deteriorating health. Immediate access to medical support can greatly reduce repercussions. Hence, there is an increasing interest in automated fall detection, often incorporated into a smart health-care system to provide better monitoring. Existing systems focus on wearable devices that are inconvenient or video monitoring that has privacy concerns. Moreover, these systems provide a limited perspective of their generalization ability as they are tested on datasets containing few activities that have wide disparity in the action space and are easy to differentiate. Complex daily life scenarios pose much greater challenges with activities that overlap in action spaces due to similar posture or motion. To overcome these limitations, we propose a fall detection model, called structure-aware discriminative feature aggregation, based on human skeletons extracted from low-resolution videos. The use of skeleton data ensures privacy and low-resolution videos ensures low hardware and computational cost. Our model captures discriminative structural displacements and motion trends using unified joint and motion features projected onto a shared high-dimensional space. Particularly, the use of separable convolution combined with a powerful graph convolutional network architecture provides improved performance. Extensive experiments on five large-scale datasets with a wide range of evaluation settings show that our model achieves competitive performance with extremely low computational complexity and runs faster than existing models.
- 10 November 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 19 (8) , 8713-8721
SDFA: Structure-Aware Discriminative Feature Aggregation for Efficient Human Fall Detection in Video
Older people are susceptible to fall due to instability in posture and deteriorating health. Immediate access to medical support can greatly reduce repercussions. Hence, there is an increasing interest in automated fall detection, often incorporated into a smart health-care system to provide better monitoring. Existing systems focus on wearable devices that are inconvenient or video monitoring that has privacy concerns. Moreover, these systems provide a limited perspective of their generalization ability as they are tested on datasets containing few activities that have wide disparity in the action space and are easy to differentiate. Complex daily life scenarios pose much greater challenges with activities that overlap in action spaces due to similar posture or motion. To overcome these limitations, we propose a fall detection model, called structure-aware discriminative feature aggregation, based on human skeletons extracted from low-resolution videos. The use of skeleton data ensures privacy and low-resolution videos ensures low hardware and computational cost. Our model captures discriminative structural displacements and motion trends using unified joint and motion features projected onto a shared high-dimensional space. Particularly, the use of separable convolution combined with a powerful graph convolutional network architecture provides improved performance. Extensive experiments on five large-scale datasets with a wide range of evaluation settings show that our model achieves competitive performance with extremely low computational complexity and runs faster than existing models.
- 1 February 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Rhythmic Component Extraction from EEG Signal Using Data Adaptive Multiband Filtering
The rhythmic components of electroencephalography (EEG) signal illustrate the neural activities and effective to implement brain computer interface (BCI). This paper presents an efficient rhythmic component extraction (RCE) method from multi-channel EEG. It is based multivariate empirical mode decomposition (MEMD) which decomposes multichannel EEG signal into a finite set of subbands termed as intrinsic mode functions (IMFs). The decomposition is fully data adaptive and effective for non-stationary signal. The IMF is a time varying band limited signal. It is filtered using a Fourier transform based zero phase bandpass filter for a specific rhythmic component. Thus obtained rhythmic component of all IMFs of an EEG channel are summed up yielding the channel's component. All the desired components of individual channels are extracted using the same method. The energies of different extracted rhythmic components are compared as a function of channels. The inter-channel correlation is taken into consideration during decomposition with MEMD hence it is very much effective for RCE from multichannel EEG signal.
- 1 February 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Rhythmic Component Extraction from EEG Signal Using Data Adaptive Multiband Filtering
The rhythmic components of electroencephalography (EEG) signal illustrate the neural activities and effective to implement brain computer interface (BCI). This paper presents an efficient rhythmic component extraction (RCE) method from multi-channel EEG. It is based multivariate empirical mode decomposition (MEMD) which decomposes multichannel EEG signal into a finite set of subbands termed as intrinsic mode functions (IMFs). The decomposition is fully data adaptive and effective for non-stationary signal. The IMF is a time varying band limited signal. It is filtered using a Fourier transform based zero phase bandpass filter for a specific rhythmic component. Thus obtained rhythmic component of all IMFs of an EEG channel are summed up yielding the channel's component. All the desired components of individual channels are extracted using the same method. The energies of different extracted rhythmic components are compared as a function of channels. The inter-channel correlation is taken into consideration during decomposition with MEMD hence it is very much effective for RCE from multichannel EEG signal.
- 1 December 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Rhythmic component retrieval from EOG artifact of electroencephalography signals
Electroencephalography (EEG) signal has been widely used to recognize the intention and cognition in brain computer interface (BCI) applications. Non-Cerebral signals like electro-oculogram (EOG) artifacts with higher energy suppress important low energy EEG components. It is required to completely remove these artifacts without loss of EEG data. This paper presents multivariate empirical mode decomposition (MEMD) based filtering approach to suppress the unwanted artifacts. The multichannel EEG signal is decomposed into a finite number of band limited signals called intrinsic mode functions (IMFs). A subband thresholding is implemented to separate artifacts using the IMFs with fractional Gaussian noise (fGn) as reference signal. Thus separated artifact includes some EEG components. Bandpass filtering is implemented to extract the EEG signal from artifact obtained by energy based thresholding. The experimental results conducted with real EEG signals show the effectiveness of the proposed method.
- 1 December 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Rhythmic component retrieval from EOG artifact of electroencephalography signals
Electroencephalography (EEG) signal has been widely used to recognize the intention and cognition in brain computer interface (BCI) applications. Non-Cerebral signals like electro-oculogram (EOG) artifacts with higher energy suppress important low energy EEG components. It is required to completely remove these artifacts without loss of EEG data. This paper presents multivariate empirical mode decomposition (MEMD) based filtering approach to suppress the unwanted artifacts. The multichannel EEG signal is decomposed into a finite number of band limited signals called intrinsic mode functions (IMFs). A subband thresholding is implemented to separate artifacts using the IMFs with fractional Gaussian noise (fGn) as reference signal. Thus separated artifact includes some EEG components. Bandpass filtering is implemented to extract the EEG signal from artifact obtained by energy based thresholding. The experimental results conducted with real EEG signals show the effectiveness of the proposed method.
- 1 December 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Removing EOG artifacts from EEG signal using noise-assisted multivariate empirical mode decomposition
Electroencephalogram (EEG) has significant applications on medical diagnosis and Brain Computer Interfacing (BCI). But the main obstacle of analyzing EEG signal is various types of noises to get actual information. Electro-oculogram (EOG) is a vital noise in EEG signal that can be produced by eye movements. De-noising EOG from EEG signal is the key issue in this research. Many research has been done on this purpose mainly Independent Component Analysis (ICA) based EOG separation with reference signal and wavelet based EOG separation. In this research, multivariate Fractional Gaussian noise channel will be used to establish a uniformly distributed reference scale and to derive the energy based threshold to detect the low frequency trends caused by EOG artifact. Avoiding these artifacts, we can get EOG free EEG hoping better results than existing methods.
- 1 December 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Removing EOG artifacts from EEG signal using noise-assisted multivariate empirical mode decomposition
Electroencephalogram (EEG) has significant applications on medical diagnosis and Brain Computer Interfacing (BCI). But the main obstacle of analyzing EEG signal is various types of noises to get actual information. Electro-oculogram (EOG) is a vital noise in EEG signal that can be produced by eye movements. De-noising EOG from EEG signal is the key issue in this research. Many research has been done on this purpose mainly Independent Component Analysis (ICA) based EOG separation with reference signal and wavelet based EOG separation. In this research, multivariate Fractional Gaussian noise channel will be used to establish a uniformly distributed reference scale and to derive the energy based threshold to detect the low frequency trends caused by EOG artifact. Avoiding these artifacts, we can get EOG free EEG hoping better results than existing methods.