Searcharxiv⌕ Search

arXiv subjects

Shehroz S. Khan

Publications and source records attributed to Shehroz S. Khan.

At least 37 records · Page 2Linked to original sources
arXiv2024

Supervised and Unsupervised Deep Learning Approaches for EEG Seizure Prediction

Epilepsy affects more than 50 million people worldwide, making it one of the world's most prevalent neurological diseases. The main symptom of epilepsy is seizures, which occur abruptly and can cause serious injury or death. The ability to predict the occurrence of an epileptic seizure could alleviate many risks and stresses people with epilepsy face. We formulate the problem of detecting preictal (or pre-seizure) with reference to normal EEG as a precursor to incoming seizure. To this end, we developed several supervised deep learning approaches to identify preictal EEG from normal EEG. We further develop novel unsupervised deep learning approaches to train the models on only normal EEG, and detecting pre-seizure EEG as an anomalous event. These deep learning models were trained and evaluated on two large EEG seizure datasets in a person-specific manner. We found that both supervised and unsupervised approaches are feasible; however, their performance varies depending on the patient, approach and architecture. This new line of research has the potential to develop therapeutic interventions and save human lives.

eess.SP↗
arXiv2024

Skeletal Video Anomaly Detection using Deep Learning: Survey, Challenges and Future Directions

The existing methods for video anomaly detection mostly utilize videos containing identifiable facial and appearance-based features. The use of videos with identifiable faces raises privacy concerns, especially when used in a hospital or community-based setting. Appearance-based features can also be sensitive to pixel-based noise, straining the anomaly detection methods to model the changes in the background and making it difficult to focus on the actions of humans in the foreground. Structural information in the form of skeletons describing the human motion in the videos is privacy-protecting and can overcome some of the problems posed by appearance-based features. In this paper, we present a survey of privacy-protecting deep learning anomaly detection methods using skeletons extracted from videos. We present a novel taxonomy of algorithms based on the various learning approaches. We conclude that skeleton-based approaches for anomaly detection can be a plausible privacy-protecting alternative for video anomaly detection. Lastly, we identify major open research questions and provide guidelines to address them.

cs.CV↗
arXiv2023

Domain-Specific Deep Learning Feature Extractor for Diabetic Foot Ulcer Detection

Diabetic Foot Ulcer (DFU) is a condition requiring constant monitoring and evaluations for treatment. DFU patient population is on the rise and will soon outpace the available health resources. Autonomous monitoring and evaluation of DFU wounds is a much-needed area in health care. In this paper, we evaluate and identify the most accurate feature extractor that is the core basis for developing a deep-learning wound detection network. For the evaluation, we used mAP and F1-score on the publicly available DFU2020 dataset. A combination of UNet and EfficientNetb3 feature extractor resulted in the best evaluation among the 14 networks compared. UNet and Efficientnetb3 can be used as the classifier in the development of a comprehensive DFU domain-specific autonomous wound detection pipeline.

cs.CV↗
arXiv2023

Temporal Shift -- Multi-Objective Loss Function for Improved Anomaly Fall Detection

Falls are a major cause of injuries and deaths among older adults worldwide. Accurate fall detection can help reduce potential injuries and additional health complications. Different types of video modalities can be used in a home setting to detect falls, including RGB, Infrared, and Thermal cameras. Anomaly detection frameworks using autoencoders and their variants can be used for fall detection due to the data imbalance that arises from the rarity and diversity of falls. However, the use of reconstruction error in autoencoders can limit the application of networks' structures that propagate information. In this paper, we propose a new multi-objective loss function called Temporal Shift, which aims to predict both future and reconstructed frames within a window of sequential frames. The proposed loss function is evaluated on a semi-naturalistic fall detection dataset containing multiple camera modalities. The autoencoders were trained on normal activities of daily living (ADL) performed by older adults and tested on ADLs and falls performed by young adults. Temporal shift shows significant improvement to a baseline 3D Convolutional autoencoder, an attention U-Net CAE, and a multi-modal neural network. The greatest improvement was observed in an attention U-Net model improving by 0.20 AUC ROC for a single camera when compared to reconstruction alone. With significant improvement across different models, this approach has the potential to be widely adopted and improve anomaly detection capabilities in other settings besides fall detection.

cs.CV↗
arXiv2023

Synthesizing Diabetic Foot Ulcer Images with Diffusion Model

Diabetic Foot Ulcer (DFU) is a serious skin wound requiring specialized care. However, real DFU datasets are limited, hindering clinical training and research activities. In recent years, generative adversarial networks and diffusion models have emerged as powerful tools for generating synthetic images with remarkable realism and diversity in many applications. This paper explores the potential of diffusion models for synthesizing DFU images and evaluates their authenticity through expert clinician assessments. Additionally, evaluation metrics such as Frechet Inception Distance (FID) and Kernel Inception Distance (KID) are examined to assess the quality of the synthetic DFU images. A dataset of 2,000 DFU images is used for training the diffusion model, and the synthetic images are generated by applying diffusion processes. The results indicate that the diffusion model successfully synthesizes visually indistinguishable DFU images. 70% of the time, clinicians marked synthetic DFU images as real DFUs. However, clinicians demonstrate higher unanimous confidence in rating real images than synthetic ones. The study also reveals that FID and KID metrics do not significantly align with clinicians' assessments, suggesting alternative evaluation approaches are needed. The findings highlight the potential of diffusion models for generating synthetic DFU images and their impact on medical training programs and research in wound detection and classification.

eess.IV↗
arXiv2023

Undersampling and Cumulative Class Re-decision Methods to Improve Detection of Agitation in People with Dementia

Agitation is one of the most prevalent symptoms in people with dementia (PwD) that can place themselves and the caregiver's safety at risk. Developing objective agitation detection approaches is important to support health and safety of PwD living in a residential setting. In a previous study, we collected multimodal wearable sensor data from 17 participants for 600 days and developed machine learning models for detecting agitation in one-minute windows. However, there are significant limitations in the dataset, such as imbalance problem and potential imprecise labelsas the occurrence of agitation is much rarer in comparison to the normal behaviours. In this paper, we first implemented different undersampling methods to eliminate the imbalance problem, and came to the conclusion that only 20% of normal behaviour data were adequate to train a competitive agitation detection model. Then, we designed a weighted undersampling method to evaluate the manual labeling mechanism given the ambiguous time interval assumption. After that, the postprocessing method of cumulative class re-decision (CCR) was proposed based on the historical sequential information and continuity characteristic of agitation, improving the decision-making performance for the potential application of agitation detection system. The results showed that a combination of undersampling and CCR improved F1-score and other metrics to varying degrees with less training time and data.

cs.LG↗
arXiv2023

Cross-Modal Video to Body-joints Augmentation for Rehabilitation Exercise Quality Assessment

Exercise-based rehabilitation programs have been shown to enhance quality of life and reduce mortality and rehospitalizations. AI-driven virtual rehabilitation programs enable patients to complete exercises independently at home while AI algorithms can analyze exercise data to provide feedback to patients and report their progress to clinicians. This paper introduces a novel approach to assessing the quality of rehabilitation exercises using RGB video. Sequences of skeletal body joints are extracted from consecutive RGB video frames and analyzed by many-to-one sequential neural networks to evaluate exercise quality. Existing datasets for exercise rehabilitation lack adequate samples for training deep sequential neural networks to generalize effectively. A cross-modal data augmentation approach is proposed to resolve this problem. Visual augmentation techniques are applied to video data, and body joints extracted from the resulting augmented videos are used for training sequential neural networks. Extensive experiments conducted on the KInematic assessment of MOvement and clinical scores for remote monitoring of physical REhabilitation (KIMORE) dataset, demonstrate the superiority of the proposed method over previous baseline approaches. The ablation study highlights a significant enhancement in exercise quality assessment following cross-modal augmentation.

cs.CV↗
arXiv2023

Rehabilitation Exercise Repetition Segmentation and Counting using Skeletal Body Joints

Physical exercise is an essential component of rehabilitation programs that improve quality of life and reduce mortality and re-hospitalization rates. In AI-driven virtual rehabilitation programs, patients complete their exercises independently at home, while AI algorithms analyze the exercise data to provide feedback to patients and report their progress to clinicians. To analyze exercise data, the first step is to segment it into consecutive repetitions. There has been a significant amount of research performed on segmenting and counting the repetitive activities of healthy individuals using raw video data, which raises concerns regarding privacy and is computationally intensive. Previous research on patients' rehabilitation exercise segmentation relied on data collected by multiple wearable sensors, which are difficult to use at home by rehabilitation patients. Compared to healthy individuals, segmenting and counting exercise repetitions in patients is more challenging because of the irregular repetition duration and the variation between repetitions. This paper presents a novel approach for segmenting and counting the repetitions of rehabilitation exercises performed by patients, based on their skeletal body joints. Skeletal body joints can be acquired through depth cameras or computer vision techniques applied to RGB videos of patients. Various sequential neural networks are designed to analyze the sequences of skeletal body joints and perform repetition segmentation and counting. Extensive experiments on three publicly available rehabilitation exercise datasets, KIMORE, UI-PRMD, and IntelliRehabDS, demonstrate the superiority of the proposed method compared to previous methods. The proposed method enables accurate exercise analysis while preserving privacy, facilitating the effective delivery of virtual rehabilitation programs.

cs.CV↗
arXiv2023

Detecting Disengagement in Virtual Learning as an Anomaly using Temporal Convolutional Network Autoencoder

Student engagement is an important factor in meeting the goals of virtual learning programs. Automatic measurement of student engagement provides helpful information for instructors to meet learning program objectives and individualize program delivery. Many existing approaches solve video-based engagement measurement using the traditional frameworks of binary classification (classifying video snippets into engaged or disengaged classes), multi-class classification (classifying video snippets into multiple classes corresponding to different levels of engagement), or regression (estimating a continuous value corresponding to the level of engagement). However, we observe that while the engagement behaviour is mostly well-defined (e.g., focused, not distracted), disengagement can be expressed in various ways. In addition, in some cases, the data for disengaged classes may not be sufficient to train generalizable binary or multi-class classifiers. To handle this situation, in this paper, for the first time, we formulate detecting disengagement in virtual learning as an anomaly detection problem. We design various autoencoders, including temporal convolutional network autoencoder, long-short-term memory autoencoder, and feedforward autoencoder using different behavioral and affect features for video-based student disengagement detection. The result of our experiments on two publicly available student engagement datasets, DAiSEE and EmotiW, shows the superiority of the proposed approach for disengagement detection as an anomaly compared to binary classifiers for classifying videos into engaged versus disengaged classes (with an average improvement of 9% on the area under the curve of the receiver operating characteristic curve and 22% on the area under the curve of the precision-recall curve).

cs.CV↗
arXiv2023

Privacy-Protecting Behaviours of Risk Detection in People with Dementia using Videos

People living with dementia often exhibit behavioural and psychological symptoms of dementia that can put their and others' safety at risk. Existing video surveillance systems in long-term care facilities can be used to monitor such behaviours of risk to alert the staff to prevent potential injuries or death in some cases. However, these behaviours of risk events are heterogeneous and infrequent in comparison to normal events. Moreover, analyzing raw videos can also raise privacy concerns. In this paper, we present two novel privacy-protecting video-based anomaly detection approaches to detect behaviours of risks in people with dementia. We either extracted body pose information as skeletons or used semantic segmentation masks to replace multiple humans in the scene with their semantic boundaries. Our work differs from most existing approaches for video anomaly detection that focus on appearance-based features, which can put the privacy of a person at risk and is also susceptible to pixel-based noise, including illumination and viewing direction. We used anonymized videos of normal activities to train customized spatio-temporal convolutional autoencoders and identify behaviours of risk as anomalies. We showed our results on a real-world study conducted in a dementia care unit with patients with dementia, containing approximately 21 hours of normal activities data for training and 9 hours of data containing normal and behaviours of risk events for testing. We compared our approaches with the original RGB videos and obtained a similar area under the receiver operating characteristic curve performance of 0.807 for the skeleton-based approach and 0.823 for the segmentation mask-based approach.

cs.CV↗
arXiv2023

Bag of States: A Non-sequential Approach to Video-based Engagement Measurement

Automatic measurement of student engagement provides helpful information for instructors to meet learning program objectives and individualize program delivery. Students' behavioral and emotional states need to be analyzed at fine-grained time scales in order to measure their level of engagement. Many existing approaches have developed sequential and spatiotemporal models, such as recurrent neural networks, temporal convolutional networks, and three-dimensional convolutional neural networks, for measuring student engagement from videos. These models are trained to incorporate the order of behavioral and emotional states of students into video analysis and output their level of engagement. In this paper, backed by educational psychology, we question the necessity of modeling the order of behavioral and emotional states of students in measuring their engagement. We develop bag-of-words-based models in which only the occurrence of behavioral and emotional states of students is modeled and analyzed and not the order in which they occur. Behavioral and affective features are extracted from videos and analyzed by the proposed models to determine the level of engagement in an ordinal-output classification setting. Compared to the existing sequential and spatiotemporal approaches for engagement measurement, the proposed non-sequential approach improves the state-of-the-art results. According to experimental results, our method significantly improved engagement level classification accuracy on the IIITB Online SE dataset by 26% compared to sequential models and achieved engagement level classification accuracy as high as 66.58% on the DAiSEE student engagement dataset.

cs.HC↗
arXiv2023

Inconsistencies in the Definition and Annotation of Student Engagement in Virtual Learning Datasets: A Critical Review

Background: Student engagement (SE) in virtual learning can have a major impact on meeting learning objectives and program dropout risks. Developing Artificial Intelligence (AI) models for automatic SE measurement requires annotated datasets. However, existing SE datasets suffer from inconsistent definitions and annotation protocols mostly unaligned with the definition of SE in educational psychology. This issue could be misleading in developing generalizable AI models and make it hard to compare the performance of these models developed on different datasets. The objective of this critical review was to explore the existing SE datasets and highlight inconsistencies in terms of differing engagement definitions and annotation protocols. Methods: Several academic databases were searched for publications introducing new SE datasets. The datasets containing students' single- or multi-modal data in online or offline computer-based virtual learning sessions were included. The definition and annotation of SE in the existing datasets were analyzed based on our defined seven dimensions of engagement annotation: sources, data modalities, timing, temporal resolution, level of abstraction, combination, and quantification. Results: Thirty SE measurement datasets met the inclusion criteria. The reviewed SE datasets used very diverse and inconsistent definitions and annotation protocols. Unexpectedly, very few of the reviewed datasets used existing psychometrically validated scales in their definition of SE. Discussion: The inconsistent definition and annotation of SE are problematic for research on developing comparable AI models for automatic SE measurement. Some of the existing SE definitions and protocols in settings other than virtual learning that have the potential to be used in virtual learning are introduced.

cs.HC↗
arXiv2022

Step Counting with Attention-based LSTM

Physical activity is recognized as an essential component of overall health. One measure of physical activity, the step count, is well known as a predictor of long-term morbidity and mortality. Step Counting (SC) is the automated counting of the number of steps an individual takes over a specified period of time and space. Due to the ubiquity of smartphones and smartwatches, most current SC approaches rely on the built-in accelerometer sensors on these devices. The sensor signals are analyzed as multivariate time series, and the number of steps is calculated through a variety of approaches, such as time-domain, frequency-domain, machine-learning, and deep-learning approaches. Most of the existing approaches rely on dividing the input signal into windows, detecting steps in each window, and summing the detected steps. However, these approaches require the determination of multiple parameters, including the window size. Furthermore, most of the existing deep-learning SC approaches require ground-truth labels for every single step, which can be arduous and time-consuming to annotate. To circumvent these requirements, we present a novel SC approach utilizing many-to-one attention-based LSTM. With the proposed LSTM network, SC is solved as a regression problem, taking the entire sensor signal as input and the step count as the output. The analysis shows that the attention-based LSTM automatically learned the pattern of steps even in the absence of ground-truth labels. The experimental results on three publicly available SC datasets demonstrate that the proposed method successfully counts the number of steps with low values of mean absolute error and high values of SC accuracy.

cs.CV↗
arXiv2022

FedSL: Federated Split Learning on Distributed Sequential Data in Recurrent Neural Networks

Federated Learning (FL) and Split Learning (SL) are privacy-preserving Machine-Learning (ML) techniques that enable training ML models over data distributed among clients without requiring direct access to their raw data. Existing FL and SL approaches work on horizontally or vertically partitioned data and cannot handle sequentially partitioned data where segments of multiple-segment sequential data are distributed across clients. In this paper, we propose a novel federated split learning framework, FedSL, to train models on distributed sequential data. The most common ML models to train on sequential data are Recurrent Neural Networks (RNNs). Since the proposed framework is privacy-preserving, segments of multiple-segment sequential data cannot be shared between clients or between clients and server. To circumvent this limitation, we propose a novel SL approach tailored for RNNs. A RNN is split into sub-networks, and each sub-network is trained on one client containing single segments of multiple-segment training sequences. During local training, the sub-networks on different clients communicate with each other to capture latent dependencies between consecutive segments of multiple-segment sequential data on different clients, but without sharing raw data or complete model parameters. After training local sub-networks with local sequential data segments, all clients send their sub-networks to a federated server where sub-networks are aggregated to generate a global model. The experimental results on simulated and real-world datasets demonstrate that the proposed method successfully trains models on distributed sequential data, while preserving privacy, and outperforms previous FL and centralized learning approaches in terms of achieving higher accuracy in fewer communication rounds.

cs.LG↗
arXiv2022

MAISON -- Multimodal AI-based Sensor platform for Older Individuals

There is a global aging population requiring the need for the right tools that can enable older adults' greater independence and the ability to age at home, as well as assist healthcare workers. It is feasible to achieve this objective by building predictive models that assist healthcare workers in monitoring and analyzing older adults' behavioral, functional, and psychological data. To develop such models, a large amount of multimodal sensor data is typically required. In this paper, we propose MAISON, a scalable cloud-based platform of commercially available smart devices capable of collecting desired multimodal sensor data from older adults and patients living in their own homes. The MAISON platform is novel due to its ability to collect a greater variety of data modalities than the existing platforms, as well as its new features that result in seamless data collection and ease of use for older adults who may not be digitally literate. We demonstrated the feasibility of the MAISON platform with two older adults discharged home from a large rehabilitation center. The results indicate that the MAISON platform was able to collect and store sensor data in a cloud without functional glitches or performance degradation. This paper will also discuss the challenges faced during the development of the platform and data collection in the homes of older adults. MAISON is a novel platform designed to collect multimodal data and facilitate the development of predictive models for detecting key health indicators, including social isolation, depression, and functional decline, and is feasible to use with older adults in the community.

cs.LG↗
arXiv2022

Multi Visual Modality Fall Detection Dataset

Falls are one of the leading cause of injury-related deaths among the elderly worldwide. Effective detection of falls can reduce the risk of complications and injuries. Fall detection can be performed using wearable devices or ambient sensors; these methods may struggle with user compliance issues or false alarms. Video cameras provide a passive alternative; however, regular RGB cameras are impacted by changing lighting conditions and privacy concerns. From a machine learning perspective, developing an effective fall detection system is challenging because of the rarity and variability of falls. Many existing fall detection datasets lack important real-world considerations, such as varied lighting, continuous activities of daily living (ADLs), and camera placement. The lack of these considerations makes it difficult to develop predictive models that can operate effectively in the real world. To address these limitations, we introduce a novel multi-modality dataset (MUVIM) that contains four visual modalities: infra-red, depth, RGB and thermal cameras. These modalities offer benefits such as obfuscated facial features and improved performance in low-light conditions. We formulated fall detection as an anomaly detection problem, in which a customized spatio-temporal convolutional autoencoder was trained only on ADLs so that a fall would increase the reconstruction error. Our results showed that infra-red cameras provided the highest level of performance (AUC ROC=0.94), followed by thermal (AUC ROC=0.87), depth (AUC ROC=0.86) and RGB (AUC ROC=0.83). This research provides a unique opportunity to analyze the utility of camera modalities in detecting falls in a home setting while balancing performance, passiveness, and privacy.

cs.CV↗
arXiv2022

Supervised Contrastive Learning for Detecting Anomalous Driving Behaviours from Multimodal Videos

Distracted driving is one of the major reasons for vehicle accidents. Therefore, detecting distracted driving behaviors is of paramount importance to reduce the millions of deaths and injuries occurring worldwide. Distracted or anomalous driving behaviors are deviations from 'normal' driving that need to be identified correctly to alert the driver. However, these driving behaviors do not comprise one specific type of driving style and their distribution can be different during the training and test phases of a classifier. We formulate this problem as a supervised contrastive learning approach to learn a visual representation to detect normal, and seen and unseen anomalous driving behaviors. We made a change to the standard contrastive loss function to adjust the similarity of negative pairs to aid the optimization. Normally, in a (self) supervised contrastive framework, the projection head layers are omitted during the test phase as the encoding layers are considered to contain general visual representative information. However, we assert that for a video-based supervised contrastive learning task, including a projection head can be beneficial. We showed our results on a driver anomaly detection dataset that contains 783 minutes of video recordings of normal and anomalous driving behaviors of 31 drivers from the various top and front cameras (both depth and infrared). Out of 9 video modalities combinations, our proposed contrastive approach improved the ROC AUC on 6 in comparison to the baseline models (from 4.23% to 8.91% for different modalities). We performed statistical tests that showed evidence that our proposed method performs better than the baseline contrastive learning setup. Finally, the results showed that the fusion of depth and infrared modalities from the top and front views achieved the best AUC ROC of 0.9738 and AUC PR of 0.9772.

cs.CV↗
arXiv2021

Improving state-of-the-art in Detecting Student Engagement with Resnet and TCN Hybrid Network

Automatic detection of students' engagement in online learning settings is a key element to improve the quality of learning and to deliver personalized learning materials to them. Varying levels of engagement exhibited by students in an online classroom is an affective behavior that takes place over space and time. Therefore, we formulate detecting levels of students' engagement from videos as a spatio-temporal classification problem. In this paper, we present a novel end-to-end Residual Network (ResNet) and Temporal Convolutional Network (TCN) hybrid neural network architecture for students' engagement level detection in videos. The 2D ResNet extracts spatial features from consecutive video frames, and the TCN analyzes the temporal changes in video frames to detect the level of engagement. The spatial and temporal arms of the hybrid network are jointly trained on raw video frames of a large publicly available students' engagement detection dataset, DAiSEE. We compared our method with several competing students' engagement detection methods on this dataset. The ResNet+TCN architecture outperforms all other studied methods, improves the state-of-the-art engagement level detection accuracy, and sets a new baseline for future research.

cs.CV↗