No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Signal Processing and Computer Vision, in Electrical and Electronic Engineering: Prospects and Challenges
Abstract
Now-a-days, signal processing is ubiquitous.
This broad electrical engineering discipline is concerned with
extracting, manipulating, and storing information embedded
in complex signals and images. From the early days of the
FFT to today’s machine/computer vision industry, signal
processing has driven many of the products and devices that
have benefited society. This chapter will inform the readers
about the current strength of the department in terms of
curriculum and research activities in this field, the contribution
of the department to society, global trends and future research
directions in this field and finally, the measures that need to be
taken to meet the upcoming goals and challenges.
ResearchGate has not been able to resolve any citations for this publication.
We propose the asynchronous control of anisotropic diffusion (AD) algorithm, and such asynchronous anisotropic diffusion (AAD) algorithm is demonstrated experimentally to reduce noise from the sensing signals obtained from Brillouin distributed optical fiber sensors. The performance of the proposed AAD algorithm is analyzed in detail for different experimental conditions and compared with that of block-matching and 3D filtering, two-dimensional wavelet denoising, AD, and non-local means algorithms. Some key factors of the proposed algorithm, such as the impact of convolution kernel size on the performance of AD algorithms, the influence of low sampling point number (SPN) on the quality of Brillouin frequency shift and the selection of diffusion thresholds are analyzed and discussed with experimental results. The experimental results validate that the AAD algorithm can provide better root-mean-square error (RMSE) and spatial resolution (SR) than the other four algorithms, especially for higher signal-to-noise ratio (SNR) improvement and higher SPNs. For lower SPNs, the performance of AAD is also not inferior to the RMSE performance of NLM and AD. The runtime of the AAD algorithm is also quite low. Moreover, the proposed algorithm offers the best SR performance as compared to other noise reduction algorithms investigated in this study. Thus, the proposed AAD algorithm can be an effective candidate to improve the measurement accuracy of Brillouin distributed optical fiber sensors.
Nurse care activity recognition is an emerging segment in healthcare automation systems based on physical movement recognition
applying machine learning techniques using various sensor-based
datasets. In this paper, different machine learning models have
been used to recognize the activities. However, before that, our user
dataset has been preprocessed using data cleaning, resampling, data
labeling, windowing, and filtering techniques in order to handle the
ununiform data. Various analytical features have been extracted
using Fast Fourier Transformation, Power Spectral Density, and Discrete Wavelet Transformation. After that, the best combinational
features have been selected from the extracted features, and class
imbalance has been mitigated before applying the conventional
machine learning models. After applying all methodology, 87.00%
accuracy has been obtained using the Light Gradient Boosting Machine Classifier.
Sussex-Huawei Locomotion-Transportation (SHL) recognition challenge, the fourth edition of the challenge provides an opportunity to recognize 8 modes of locomotion and transportation (activities) from radio data, including GPS reception, GPS location, WiFi reception, and GSM cell tower scans in a user-independent manner. Though the dataset has provided good location data, the main challenge has been how to use the variable number of data from satellite, WiFi, and cell tower information. We, team “Guerrilla”, have used the location data to extract different features like distance, velocity, acceleration, and hand-picked certain information from the satellite, WiFi, and cell tower data. Various features have been extracted from the frequency and time domain and diverse windowing techniques have been implemented. Using 1350 features, different models like Extra Trees Classifier, Random Forest Classifier, XGBoost Classifier, etc. have been trained among which the XGBoost Classifier model has performed the best. And finally, the top 250 features have been selected for the XGBoost Classifier to build an efficient model that has given accuracy and f1 score of 79.11% and 79.39% respectively on the validation dataset.
Recognition of daily human activities in various locomotion and transportation modes has numerous applications like coaching users for behavior modification and maintaining a healthy lifestyle. Besides, applications and user interfaces aware of user mobility through their smartphones can also aid in urban transportation planning, smart parking, and vehicular traffic monitoring. In this paper, we explored smartphone sensor-based two benchmark datasets (Sussex Huawei Locomotion (SHL) and Transportation Mode Detection (TMD)). Firstly, we demonstrated preprocesssing of sensor data, window length optimization based on Akaike Information Criteria (AIC), and introduced smartphone orientation independent features. We also provided an in-depth analysis of different smartphone sensors’ importance for classifying daily activities and transportation modes. We justified the sensor relevance by showing the variation of performances with the number of sensors explored. For refining classifier predictions, we also proposed a post-processing approach named “Mode technique”. This method primarily concentrates on the statistical analysis of transportation modes and improves the activity recognition rate in statistical classifiers: Decision Tree, K-Nearest Neighbors, Linear Discriminant Analysis, Logistic Regression, Support Vectors Machine with RBF kernel, Random Forest, and deep learning-based methods: Artificial Neural Network and Recurrent Neural Network by smoothing the outputs of these classifiers. Besides, we showed the use of magnitude and jerk-based features to overcome the overfitting problem due to smartphone orientation. We obtained 97.2% accuracy in the SHL dataset and 99.13% accuracy in the TMD dataset. These results demonstrate that our approach can profoundly recognize various activities in advanced locomotion and transportation modes compared to existing methods in two large-scale datasets.
Abstract Background/ Introduction: Autism Spectrum Disorder (ASD) is a neuro-developmental disorder that limits social and cognitive abilities. ASD has no cure so early diagnosis is important for reducing its impact. The current behavioural observation-based subjective-diagnosis systems (e.g., DSM-5 or ICD-10) frequently misdiagnose subjects. Therefore, researchers are attempting to develop automated diagnosis-systems with minimal human intervention, quicker screening time, and better outreach.
Method: This paper is a PRISMA-based systematic review examining the potential
of automated autism detection system with Human Activity Analysis (HAA) to look for distinctive ASD-characteristics such as repetitive behaviour, abnormal gait and visual saliency. The literature from 2011 onward is qualitatively and quantitatively analysed to investigate whether HAA can identify the features of ASD, the level of its classification accuracy, the degree of human intervention, and screening time. Based on these findings we discuss the approaches, challenges, resources, and future directions in this area.
Result: According to our quantitative assessment of the dataset[1], Inception
v3 and LSTM [1] give the highest accuracy (89%) for repetitive behavior. For
abnormal gait-based approach, the Multilayer Perceptron gives 98% accuracy
based on 18 features from dataset [2]. For gaze pattern, a saliency-metric
feature-based learning [3] gives 99% accuracy on dataset[4], while an algorithm
involving statistical features and Decision Trees yields an accuracy of 76% on
dataset [5].
Conclusion: In terms of the state-of-the-art, fully automated HAA systems
for ASD diagnosis show promise but are still in developmental stages. However,
this is an active research field, and HAA has good prospects for helping to
diagnose ASD objectively in less time with better accuracy.
Action recognition is a very widely explored research area in computer vision and related fields. We propose Kinematics Posture Feature (KPF) extraction from 3D joint positions based on skeleton data for improving the performance of action recognition. In this approach, we consider the skeleton 3D joints as kinematics sensors. We propose Linear Joint Position Feature (LJPF) and Angular Joint Position Feature (AJPF) based on 3D linear joint positions and angles between bone segments. We then combine these two kinematics features for each video frame for each action to create the KPF feature sets. These feature sets encode the variation of motion in the temporal domain as if each body joint represents kinematics position and orientation sensors. In the next stage, we process the extracted KPF feature descriptor by using a low pass filter, and segment them by using sliding windows with optimized length. This concept resembles the approach of processing kinematics sensor data. From the segmented windows, we compute the Position-based Statistical Feature (PSF). These features consist of temporal domain statistical features (e.g., mean, standard deviation, variance, etc.). These statistical features encode the variation of postures (i.e., joint positions and angles) across the video frames. For performing classification, we explore Support Vector Machine (Linear), RNN, CNNRNN, and ConvRNN model. The proposed PSF feature sets demonstrate prominent performance in both statistical machine learning- and deep learning-based models. For evaluation, we explore five benchmark datasets namely UTKinect-Action3D, Kinect Activity Recognition Dataset (KARD), MSR 3D Action Pairs, Florence 3D, and Office Activity Dataset (OAD). To prevent overfitting, we consider the leave-one-subject-out framework as the experimental setup and perform 10-fold cross-validation. Our approach outperforms several existing methods in these benchmark datasets and achieves very promising classification performance.
Nursing activity recognition adds a new dimension to the healthcare automation system. But nursing activity recognition is very challenging than identifying simple human activities like walking, cycling, swimming, etc. due to intra-class variability between activities. Besides, the lack of proper dataset does not allow researchers to develop a generalized method for nursing activity or comparing baseline methods on different datasets. Nurse Care Activity Recognition Challenge 2020 provides a dataset of twelve nursing activities. In this paper, we have described our (Team Hex Code) approach where we have emphasized on developing method, which can cope up with real-world data with noise and uncertainty. In our method, we have resampled our data to deal with a variable sample frequency of dataset and we have also applied feature selection method on the extracted feature to have the best combination of feature set for classification. We have used random forest classifier which is a classical machine learning algorithm. Applying our methodology, we have got 78% validation accuracy on the dataset. We have trained our model on the lab dataset and validate them on the field dataset.
Cooking Activity Recognition Challenge [1] is organized as a part of ABC2020 [2]. In this work, we analyze and summarize the approaches of submissions of the Challenge. A dataset consisting of macro and micro activities, collected in a Cooking scenario were opened to the public with a goal of recognizing both of these activities. The participant teams used the dataset and submitted their predictions of test data which was released on March 1st, 2020. The submission of the teams was evaluated rigorously and the winning team achieved about 92.08% averaged accuracy for macro- and micro activities.
In this paper, the characteristics of Brillouin gain spectrum (BGS) obtained from a Brillouin optical time-domain analysis (BOTDA) sensor are investigated and analyzed experimentally. The measured BGSs obtained for various pump-pulse widths and temperatures are fitted with different spectrum profiles using nonlinear least-squares curve fitting technique. The fitting performances of used profiles are presented and analyzed. Based on such performances, the proper spectrum profile to be used in the fitting process is determined and used to extract key parameters of the measured BGSs accurately. The variations of such key parameters with pump-pulse widths and temperatures are also investigated and analyzed. The results reveal that pump-pulse widths and temperatures have significant effects on the extracted key parameters of the measured BGSs obtained from BOTDA sensors. Bangladesh Journal of Physics, 27(1), 69-80, June 2020
Nurse care activity recognition is a new challenging research field in human activity recognition (HAR) because unlike other activity recognition, it has severe class imbalance problem and intra-class variability depending on both the subject and the receiver. In this paper, we applied the Random Forest-based resampling method to solve the class imbalance problem in the Heiseikai data, nurse care activity dataset. This method consists of resampling, feature selection based on Gini impurity, and model training and validation with Stratified KFold cross-validation. By implementing the Random Forest classifier, we achieved 65.9% average cross-validation accuracy in classifying 12 activities conducted by nurses in both lab and real-life settings. Our team, "Britter Baire" developed this algorithmic pipeline for "The 2nd Nurse Care Activity Recognition Challenge Using Lab and Field Data".
Brain disorder characterized by seizure is a common disease among people in the world. Characterization of electroencephalogram (EEG) signals in terms of complexity can be used to identify neurological disorders. In this study, a non-linear epileptic seizure detection method based on multiscale entropy (MSE) has been employed to characterize the complexity of EEG signals. For this reason, the MSE method has been applied on Bonn dataset containing seizure and non-seizure EEG data and the corresponding results in terms of complexity have been obtained. Using statistical tests and support vector machine (SVM), the classification ability of the MSE method has been verified on Bonn dataset. Our results show that the MSE method is a viable approach to identifying epileptic seizure demonstrating a classification accuracy of 91.7%.
Wearable sensor-based systems and devices have been expanded in different application domains, especially in the healthcare arena. Automatic age and gender estimation has several important applications. Gait has been demonstrated as a profound motion cue for various applications. A gait-based age and gender estimation challenge was launched in the 12th IAPR International Conference on Biometrics (ICB), 2019. In this competition, 18 teams initially registered from 14 countries. The goal of this challenge was to find some smart approaches to deal with age and gender estimation from sensor-based gait data. For this purpose, we employed a large wearable sensor-based gait dataset, which has 745 subjects (357 females and 388 males), from 2 to 78 years old in the training dataset; and 58 subjects (19 females and 39 males) in the test dataset. It has several walking patterns. The gait data sequences were collected from three IMUZ sensors, which were placed on waist-belt or at the top of a backpack. There were 67 solutions from ten teams—for age and gender estimation. This paper extensively analyzes the methods and achieved-results from various approaches. Based on analysis, we found that deep learning-based solutions lead the competitions compared with conventional handcrafted methods. We found that the best result achieved 24.23% prediction error for gender estimation, and 5.39 mean absolute error for age estimation by employing angle embedded gait dynamic image and temporal convolution network.
Abstract Gait-based features provide the potential for a subject to be recognized even from a low-resolution image sequence, and they can be captured at a distance without the subject’s cooperation. Person recognition using gait-based features (gait recognition) is a promising real-life application. However, several body parts of the subjects are often occluded because of beams, pillars, cars and trees, or another walking person. Therefore, gait-based features are not applicable to approaches that require an unoccluded gait image sequence. Occlusion handling is a challenging but important issue for gait recognition. In this paper, we propose silhouette sequence reconstruction from an occluded sequence (sVideo) based on a conditional deep generative adversarial network (GAN). From the reconstructed sequence, we estimate the gait cycle and extract the gait features from a one gait cycle image sequence. To regularize the training of the proposed generative network, we use adversarial loss based on triplet hinge loss incorporating Wasserstein GAN (WGAN-hinge). To the best of our knowledge, WGAN-hinge is the first adversarial loss that supervises the generator network during training by incorporating pairwise similarity ranking information. The proposed approach was evaluated on multiple challenging occlusion patterns. The experimental results demonstrate that the proposed approach outperforms the existing state-of-the-art benchmarks.
Sensor-based recognition of locomotion and transportation modes has numerous application domains including urban traffic monitoring, transportation planning, and healthcare. However, the use of a smartphone in a fixed position and orientation in previous research works limited the user behavior a lot. Besides, the performance of naive methods for position-independent cases was not up to the mark. In this research, we have designed a position and orientation independent deep ensemble network (POIDEN) to classify eight modes of locomotion and transportation activities. The proposed POIDEN architecture is constructed of a Recurrent Neural Network (RNN) with LSTM that is assigned the task of selecting optimum general classifiers (random forest, decision tree, gradient boosting, etc.) to classify the activity labels. We have trained the RNN architecture using an intermediate feature set (IFS), whereas, the general classifiers have been trained using a statistical classifier feature set (SCFS). The choice of a classifier by RNN is dependent upon the highest probability of those classifiers to recognize particular activity samples. We have also utilized the rotation of acceleration and magnetometer values from phone coordinate to earth coordinate, proposed jerk feature, and position insensitive features along with parameter adjustment to make the POIDEN architecture position and orientation independent. Our team "Gradient Descent" has presented this work for the "Sussex-Huawei Locomotion-Transportation (SHL) recognition challenge".
In this paper, we have used a smartphone sensor-based benchmark Sussex-Huawei Locomotion-Transportation (SHL) dataset for rich locomotion and transportation analytics. We have shown a comparison of different sensor-based features for the identification of a specific activity level. Besides, we have proposed a "Mod technique" method, which increases the accuracy of classifier outputs in offline processing method. We have assumed that the minimum switching time from one transportation activity to another is 1-minute as a base of this technique. Based on this assumption we have been able to correct the wrong prediction of the classifier outputs in each 1-minute data frame. Our proposed method brings 4.56% accuracy increase in the Random Forest (RnF) classifier output. Our team, "Confusion Matrix" has developed this algorithm pipeline for the "Sussex-Huawei Locomotion-Transportation (SHL) recognition challenge".
Tanmoy Paul, Ummul Afia Shammi, Mosabber Uddin Ahmed, Rashedur Rahman, Syoji Kobashi, and Md Atiqur Rahman Ahad, "A Study on Face Detection Using Viola-Jones Algorithm for Various Backgrounds, Angels and Distances", International Journal of Biomedical Soft Computing and Human Sciences, Vol. 23, No. 1, 2018.
The University of Dhaka mobility data set (DU-MD) is a human action recognition (HAR) data set consisting of 10 classes and 5000 observations from 50 subjects recorded using wrist-mounted sensors embracing accelerometry. The data set exhibits sufficient statistical diversity in physiological parameters and a noteworthy correlation between similar activities with coveted quantitative and qualitative features, suitable for training machine learning models. On the other hand, the wrist-mounted approach parallels the future commercial scenarios. In this paper, we explore how the quantitative features of the DU-MD have been extracted and selected. Existing machine learning models used in HAR, in particular, support vector machines, ensemble of classifi ers, and subspace K-nearest neighbours have been applied to our data set for activity and fall classi cation, with outcomes being compared with benchmark and similar data sets. With a HAR classi cation accuracy of 93%, fall detection accuracy of 97% and fall classi cation of 68.3%, quantitative performance metrics have either approached or outperformed other data sets, making this dataset suitable for application in hardware-independent healthcare monitoring systems. Finally, we construct an algorithm with our data set based on performance metrics, and suggest some strategies for large-scale commercial implementation.
We propose and experimentally demonstrate the use of principal component analysis (PCA) based pattern recognition to extract temperature distribution from the measured Brillouin gain spectra (BGSs) along the fiber under test (FUT) obtained by Brillouin optical time domain analysis (BOTDA) system. The proposed scheme employs a reference database consisting of relevant ideal BGSs with known temperature attributes. PCA is then applied to the BGSs in the reference database as well as to the measured BGSs so as to reduce their size by extracting their most significant features. Now, for each feature vector of the measured BGS, we determine its best match in the reference database comprised of numerous reduced-size feature vectors of the ideal BGSs. The known temperature attribute corresponding to the best-matched BGS in the reference database is then taken as the extracted temperature of the measured BGS. We analyzed the performance of PCA-based pattern recognition algorithm in detail and compared it with that of curve fitting method. The experimental results validate that the proposed technique can provide better accuracy, faster processing speed and larger noise tolerance for the measured BGSs. Therefore, the proposed PCA-based pattern recognition algorithm can be considered as an attractive method for extracting temperature distributions along the fiber in BOTDA sensors.
The recently introduced multivariate multiscale entropy (MMSE) has been successfully used to quantify structural complexity in terms of nonlinear within- and cross-channel correlations as well as to reveal complex dynamical couplings and various degrees of synchronization over multiple scales in real-world multichannel data. However, the applicability of MMSE is limited by the coarse-graining process which defines scales, as it successively reduces the data length for each scale and thus yields inaccurate and undefined entropy estimates at higher scales and for short length data. To that cause, we propose the multivariate multiscale fuzzy entropy (MMFE) algorithm and demonstrate its superiority over the MMSE on both synthetic as well as real-world uterine electromyography (EMG) short duration signals. Based on MMFE features, an improvement in the classification accuracy of term-preterm deliveries was achieved, with a maximum area under the curve (AUC) value of 0.99.
We experimentally demonstrate the use of artificial neural network (ANN) to process sensing signals obtained from Brillouin optical time domain analyzer (BOTDA). The distributed temperature information is extracted directly from the local Brillouin gain spectra (BGSs) along the fiber under test without the process of determination of Brillouin frequency shift (BFS) and hence conversion from BFS to temperature. Unlike our previous work for short sensing distance where ANN is trained by measured BGSs, here we employ ideal BGSs with different linewidths to train the ANN in order to take the linewidth variation due to different conditions from the training and testing phases into account, making it feasible for long distance sensing. Moreover, the performance of ANN is compared with other two techniques, Lorentzian curve fitting and cross-correlation method, and our results show that ANN has higher accuracy and larger tolerance to measurement error, especially at large frequency scanning step. We also show that the temperature extraction from BOTDA measurements employing ANN is significantly faster than the other two approaches. Hence ANN can be an excellent alternative tool to process BGSs measured by BOTDA and obtain temperature distribution along the fiber, especially when large frequency scanning step is adopted to significantly reduce the measurement time but without sacrifice of sensing accuracy.
A data-adaptive algorithm for the entropy-based analysis of struc- tural regularities (complexity) in multivariate signals is proposed. This is achieved by combining multivariate sample entropy with a multivariate extension of empirical mode decomposition, both data-driven multiscale techniques. The proposed analysis across data-adaptive scales makes the approach robust to nonstationarity, a critical issue with information theoretic measures. Simulations on synthetic and real-world physiological data support the approach and validate the hypothesis of increased complexity for unconstrained as compared to constrained (due to e.g. ageing or illness) biological systems.
The recently introduced Fast and Adaptive Bidimensional Empirical Mode Decomposition (FABEMD) method is evaluated in image fusion applications. The FABEMD method does not suffer from the problems related to the traditional two dimensional scattered data interpolation based EMD method. As a result, the quality of the extracted Bidimensional Intrinsic Mode Functions (BIMFs) are better and the whole process is less time-consuming, a desirable feature for real-world image fusion. Besides, the ability to generate the same number of BIMFs with matching scales and the structure recognition capability facilitate heterogeneous applications. Simulation results demonstrate the effectiveness of the proposed approach in the context of multi-focus image fusion.
This work generalizes the recently introduced univariate multiscale entropy (MSE) analysis to the multivariate case. This is achieved by introducing multivariate sample entropy (MSampEn) in a rigorous way, in order to account for both within- and cross-channel dependencies in multiple data channels, and by evaluating it over multiple temporal scales. The multivariate MSE (MMSE) method is shown to provide an assessment of the underlying dynamical richness of multichannel observations, and more degrees of freedom in the analysis than standard MSE. The benefits of the proposed approach are illustrated by simulations on complexity analysis of multivariate stochastic processes and on real-world multichannel physiological and environmental data.
Multivariate physical and biological recordings are common and their simultaneous analysis is a prerequisite for the understanding of the complexity of underlying signal generating mechanisms. Traditional entropy measures are maximized for random processes and fail to quantify inherent long-range dependencies in real world data, a key feature of complex systems. The recently introduced multiscale entropy (MSE) is a univariate method capable of detecting intrinsic correlations and has been used to measure complexity of single channel physiological signals. To generalize this method for multichannel data, we first introduce multivariate sample entropy (MSampEn) and evaluate it over multiple time scales to perform the multivariate multiscale entropy (MMSE) analysis. This makes it possible to assess structural complexity of multivariate physical or physiological systems, together with more degrees of freedom and enhanced rigor in the analysis. Simulations on both multivariate synthetic data and real world postural sway analysis support the approach.
In this paper, the loss of attention of automotive drivers is studied by using eye blink detection. Facial landmark detection for detecting eye is explored. Afterward, eye blink is detected using Eye Aspect Ratio. By comparing the time of eye closure to a particular period, the driver’s tiredness is decided. The total number of eye blinks in a minute is counted to detect drowsiness. Calculation of total eye blinks in a minute for the driver is done, then compared it with a known standard value. If any of the above conditions fulfills, the system decides the driver is unconscious. A total of 120 samples were taken by placing the light source front, back, and side. There were 40 samples for each position of the light source. The maximum error rate occurred when the light source was placed back with a 15% error rate. The best scenario was 7.5% error rate where the light source was placed front side. The eye blinking process gave an average error of 11.67% depending on the various position of the light source. Another 120 samples were taken at a different time of the day for calculating total eye blink in a minute. The maximum number of blinks was in the morning with an average blink rate of 5.78 per minute, and the lowest number of blink rate was in midnight with 3.33% blink rate. The system performed satisfactorily and achieved the eye blink pattern with 92.7% accuracy.
The continuous operation of Brillouin optical fiber distributed sensors requires the acquisition and processing of large amount of data that imposes ample pressure on the data acquisition systems and storage devices. To overcome such limitations, the compressed sensing (CS) method based on K-Singular value decomposition (K-SVD) algorithm as transform is proposed in this paper for the accurate extraction of sensing information with less data acquired from Brillouin optical fiber distributed sensors without the hardware modification. In such algorithm, the parameters of the K-SVD dictionary are selected first through simulation and the signals are sampled and reconstructed successfully. Then, we apply the regularized orthogonal matching pursuit (ROMP) reconstruction algorithm to effectively reconstruct the experimental data obtained with different sampling rates, frequency steps and test temperatures. We also compare the performances of the CS method based on K-SVD with that based on discrete cosine transform (DCT), discrete Fourier transform (DFT) and principle component analysis (PCA) as transforms. The experimental results show that the proposed method can effectively reconstruct the experimental signals with much reduced amount of data as compared to other methods. The proposed method also provides satisfactory measurement accuracy in extracting sensing information. Therefore, the K-SVD based CS method can be an attractive alternative to make Brillouin optical fiber distributed sensors more suitable for faster operation.
An indispensable part of life is social interaction. Without any doubt, it can help to overcome Autism Spectrum Disorder (ASD). ASD is defined as an abnormality in social communication, which is not a disease. So, to face this disorder, involvement in social communication is required. It can be possible through therapy sessions. In this modern era, not only the human but also a robot can play the role of the interaction partner during these sessions. To examine the behavior of the child with ASD, data is required, which is not possible to get all the time. From this thought, the DREAM dataset paves the way to evaluate the Robot Enhanced Therapy and records the data of the 61 children with ASD. The dataset provides the skeleton-based, gaze-based features, which were recorded by the RGB-D cameras. Besides, characteristics of the therapy and age, gender, ID number of a child are also provided. In this paper, we have proposed a method to classify the tasks (Imitation, Joint Attention, and Turn-Taking) accomplished by those children provided in the dataset. Further analysis is also executed by us to assess if a robot can be a substitute for humans to conduct the therapy session. Skeleton joint positions and gaze-based vectors are utilized to draw out the angles, distances between joints, and the coordinate direction angles, Direction Gaze Zone (DGZ), respectively. From the skeleton-based approach, statistical features (such as mean, median, standard deviation, minimum, and maximum), and from the gaze-based approach, mean frequency, the number of peaks of a signal of the frequency domain are obtained. Therefore, the entire analysis is evaluated from two perspectives. The ensemble methods such as Random Forest classifier, XGBoost classifier, Extra Trees classifier are deployed to obtain the predicted result from the test data. The results are satisfactory considering the challenge and complexity of the ASD domain. We have explored how robots can be an alternative to humans for the improvement of social communication among children with ASD.
https://ieeexplore.ieee.org/abstract/document/9638874
The book introduces some challenging methods and solutions to solve the human activity recognition challenge. This book highlights the challenge that will lead the researchers in academia and industry to move further related to human activity recognition and behavior analysis, concentrating on cooking challenge. Current activity recognition systems focus on recognizing either the complex label (macro-activity) or the small steps (micro-activities) but their combined recognition is critical for analysis like the challenge proposed in this book. It has 10 chapters from 13 institutes and 8 countries (Japan, USA, Switzerland, France, Slovenia, China, Bangladesh, and Columbia).
The Sussex-Huawei Locomotion-Transportation (SHL) Challenge 2020 was an open competition of recognizing eight different activities that had been performed by three individual users and participants of this competition were tasked to classify these eight different activities with modes of locomotion and transportation. This year's data was recorded with a smartphone which was located in four different body positions. The primary challenge was to make a user-invariant as well as position-invariant classification model. The train set consisted of data from only user-1 with all positions whereas the test set consisted of data from user 2 and 3 with unspeicified sensor position. Moreover, a small validation with the same charecteristics of the test set was given to validate the classifier. In this paper, we have described our (Team Red Circle) approach in which we have used previous year's challenge data as well as this year's provided data to make our training dataset and validation set that have helped us to make our model generative. In our approach, we have extracted various types of features to make our model user independent and position invariant, we have applied Random Forest classifier which is a classical machine learning algorithm and achieved 92.69% accuracy on our customized train set and 77.04% accuracy on our customized validation set.
Detection of mental stress has been receiving great attention from the researchers for many years. Many studies have analyzed electroencephalogram signals in order to estimate mental stress using linear methods. In this paper, a novel nonlinear stress assessment method based on multivariate multiscale entropy has been introduced. Since the multivariate multiscale entropy method characterizes the complexity of nonlinear time series, this research determines the mental stress of human during cognitive workload using complexity of electroencephalogram (EEG) signals. To perform this work, 36 subjects including 9 men and 27 women were participated in the cognitive workload experiment. Multivariate multiscale entropy method has been applied to electroencephalogram data collected from those subjects for estimating mental stress in terms of complexity. The complexity feature of brain electroencephalogram signals collected during resting and cognitive workload has shown statistically significant ([Formula: see text]) differences across brain regions and mental tasks which can be implemented practically for building stress detection system. In addition, the complexity profile of electroencephalogram signals has shown that higher stress is reflected in good counting compared to bad counting. Moreover, the support vector machine (SVM) has shown promising classification between resting and mental counting states by providing 80% sensitivity, 100% specificity and 90% classification accuracy.
The use of an artificial neural network (ANN) for extraction of a temperature profile from a local Brillouin gain spectrum in a Brillouin optical time-domain analysis fibre sensor system is proposed and demonstrated. An ANN is applied to process the Brillouin timedomain trace in order to extract the temperature information along the fibre after the data acquisition process. The results show that the ANN provides higher accuracy and larger tolerance to measurement error than Lorentzian curve fitting does, especially for a large frequency scanning step. Hence the measurement time can be greatly reduced by adopting a larger frequency scanning step without sacrificing accuracy.
This paper contains experimental procedure of webcam-based eye-tracker specially for low power devices. This paper consists of five processes. First one is background suppression for reducing average processing requirement. Second one is Haar-cascade feature-based face detection algorithm. Third one is geometrically eye-position determination. The fourth part is to detect and track eye-ball center using mean of gradient vector. The fifth and last step is to detect where the user looking. We simply calculate percentage of movement of eye to detect either it looking left, right, up or down. This procedure is highly effective with satisfactory accuracy. It also requires less processing power.
Display of stereo images is widely used to enhance the viewing experience of three-dimensional imaging and communication systems. In this paper, we propose a method for estimating the quality of stereoscopic images using segmented image features and disparity. This method is inspired by the human visual system. We believe the perceived distortion and disparity of any stereoscopic display is strongly dependent on local features, such as edge (non-plane) and non-edge (plane) areas. Therefore, a no-reference perceptual quality assessment is developed for JPEG coded stereoscopic images based on segmented local features of artifacts and disparity. Local feature information such as edge and non-edge area based relative disparity estimation, as well as the blockiness and the blur within the block of images are evaluated in this method. Two subjective stereo image databases are used to evaluate the performance of our method. The subjective experiments results indicate our model has sufficient prediction performance. Bibtex entry for this abstract Preferred format for this abstract (see Preferences) Find Similar Abstracts: Use: Authors Title Abstract Text Return: Query Results Return items starting with number Query Form Database: Astronomy Physics arXiv e-prints
Established complexity measures typically operate at a single scale and thus fail to quantify inherent long-range correlations in real-world data, a key feature of complex systems. The recently introduced multiscale entropy (MSE) method has the ability to detect fractal correlations and has been used successfully to assess the complexity of univariate data. However, multivariate observations are common in many real-world scenarios and a simultaneous analysis of their structural complexity is a prerequisite for the understanding of the underlying signal-generating mechanism. For this purpose, based on the notion of multivariate sample entropy, the standard MSE method is extended to the multivariate case, whereby for rigor, the intrinsic multivariate scales of the input data are generated adaptively via the multivariate empirical mode decomposition (MEMD) algorithm. This allows us to gain better understanding of the complexity of the underlying multivariate real-world process, together with more degrees of freedom and physical interpretation in the analysis. Simulations on both synthetic and real-world biological multivariate data sets support the analysis.
Perceptual image quality evaluation has become an important issue, due to increasing transmission of multimedia contents over the Internet and 3G mobile networks. Most of the no reference perceptual image quality evaluations traditionally attempted to quantify the predefined artifacts of the coded images. Under the assumption that human visual perception is very sensitive to edge information of an image and any kinds of artifacts create pixel distortion, we propose a new approach for designing a no reference image quality evaluation model for JPEG2000 images in this paper, which uses pixel distortions and edge information. Subjective experiment results on the images are used to train and test the model, which has achieved good quality prediction performance.
For system identification problems, such as noise and echo cancellation, FIR adaptive filters are mainly used for their simple adaptation and numerical stability. When the unknown system is a high-Q resonant system, having a very long impulse response, IIR adaptive filters are more efficient for reduction in the order of a transfer function. One way to realize the IIR adaptive filter is a separate form, in which the numerator and the denominator are separately realized and adjusted. In the actual applications, the order of the unknown system is not known. In this case, it is very important to estimate the total order and the order assignment on the numerator and the denominator. In this paper, effects of the order estimation error on the residual error are investigated. In this form, indirect error evaluation called `equation error' is used. Through theoretical and numerical investigation, the following results are obtained. First, under estimation of the order of the denominator causes large degradation. Second, over estimation can improve the performance. However, this improvement is saturated to some extent due to cancellation of the redundant poles and zeros. Third, the system identification error is proportional to the equation error as the adaptive filter approaching the optimum. Finally, there is possibility of recovering from the unstable state as the order assignment approaches to the optimum in an adaptive process using the equation error. Computer solutions are provided to aid in gaining insight of the order assignment and stability problem.
Adaptive noise cancellation from speech signals using SUSC algorithm
Mosabber Uddin Ahmed, Md. Shafiul Alam, A.H.M. Asadul
Huq and Farruk Ahmed, "Adaptive noise cancellation
from speech signals using SUSC algorithm", The Dhaka
University Journal of Science, vol.54, no.2, pp.181-186,
2006.
Atiqur Rahman Ahad, Computer Vision and Action
- Md
Md. Atiqur Rahman Ahad, Computer Vision and Action
Cooking Activity Recognition with Varying Sampling Rates using Deep Convolutional GRU Framework
- Sadman Md
- Omar Siraj
- M A R Shahid
- Ahad
Md. Sadman Siraj, Omar Shahid, and M.A.R. Ahad,
"Cooking Activity Recognition with Varying Sampling
Rates using Deep Convolutional GRU Framework", in
Human Activity Recognition Challenge, Springer Nature
Switzerland AG, 115-126, 2021.
Visual face scanning and emotion perception analysis between Autistic and Typically Developing children
- Ziaul Uzma Haque Syeda
- Zishan Zahidul Zafar
- Islam
- Miftahul Syed Mahir Tazwar
- Koichi Jannat Rasna
- Md Kise
- Atiqur Rahman Ahad
Uzma Haque Syeda, Ziaul Zafar, Zishan Zahidul Islam,
Syed Mahir Tazwar, Miftahul Jannat Rasna, Koichi Kise,
and Md. Atiqur Rahman Ahad, "Visual face scanning and
emotion perception analysis between Autistic and Typically
Developing children", ACM UbiComp Workshop on Mental
Health and Well-being: Sensing and Intervention, Hawaii,
USA, 2017.