RESUMO
The number of cameras utilised in smart city domains is increasingly prominent and notable for monitoring outdoor urban and rural areas such as farms and forests to deter thefts of farming machinery and livestock, as well as monitoring workers to guarantee their safety. However, anomaly detection tasks become much more challenging in environments with low-light conditions. Consequently, achieving efficient outcomes in recognising surrounding behaviours and events becomes difficult. Therefore, this research has developed a technique to enhance images captured in poor visibility. This enhancement aims to boost object detection accuracy and mitigate false positive detections. The proposed technique consists of several stages. In the first stage, features are extracted from input images. Subsequently, a classifier assigns a unique label to indicate the optimum model among multi-enhancement networks. In addition, it can distinguish scenes captured with sufficient light from low-light ones. Finally, a detection algorithm is applied to identify objects. Each task was implemented on a separate IoT-edge device, improving detection performance on the ExDark database with a nearly one-second response time across all stages.
RESUMO
The combination of deep-learning and IoT plays a significant role in modern smart solutions, providing the capability of handling task-specific real-time offline operations with improved accuracy and minimised resource consumption. This study provides a novel hardware-aware neural architecture search approach called ESC-NAS, to design and develop deep convolutional neural network architectures specifically tailored for handling raw audio inputs in environmental sound classification applications under limited computational resources. The ESC-NAS process consists of a novel cell-based neural architecture search space built with 2D convolution, batch normalization, and max pooling layers, and capable of extracting features from raw audio. A black-box Bayesian optimization search strategy explores the search space and the resulting model architectures are evaluated through hardware simulation. The models obtained from the ESC-NAS process achieved the optimal trade-off between model performance and resource consumption compared to the existing literature. The ESC-NAS models achieved accuracies of 85.78%, 81.25%, 96.25%, and 81.0% for the FSC22, UrbanSound8K, ESC-10, and ESC-50 datasets, respectively, with optimal model sizes and parameter counts for edge deployment.
RESUMO
Deep-learning models play a significant role in modern software solutions, with the capabilities of handling complex tasks, improving accuracy, automating processes, and adapting to diverse domains, eventually contributing to advancements in various industries. This study provides a comparative study on deep-learning techniques that can also be deployed on resource-constrained edge devices. As a novel contribution, we analyze the performance of seven Convolutional Neural Network models in the context of data augmentation, feature extraction, and model compression using acoustic data. The results show that the best performers can achieve an optimal trade-off between model accuracy and size when compressed with weight and filter pruning followed by 8-bit quantization. In adherence to the study workflow utilizing the forest sound dataset, MobileNet-v3-small and ACDNet achieved accuracies of 87.95% and 85.64%, respectively, while maintaining compact sizes of 243 KB and 484 KB, respectively. Henceforth, this study concludes that CNNs can be optimized and compressed to be deployed in resource-constrained edge devices for classifying forest environment sounds.
RESUMO
Internet services have collected our personal data since their inception. In the beginning, the personal data collection was uncoordinated and was limited to a few selected data types such as names, ages, birthdays, etc. Due to the widespread use of social media, more and more personal data has been collected by different online services. We increasingly see that Internet of Things (IoT) devices are also being adopted by consumers, making it possible for companies to capture personal data (including very sensitive data) with much less effort and autonomously at a very low cost. Current systems architectures aim to collect, store, and process our personal data in the cloud with very limited control when it comes to giving back to citizens. However, Personal Data Stores (PDS) have been proposed as an alternative architecture where personal data will be stored within households, giving us complete control (self-sovereignty) over our data. This paper surveys the current literature on Personal Data Stores (PDS) that enable individuals to collect, control, store, and manage their data. In particular, we provide a comprehensive review of related concepts and the expected benefits of PDS platforms. Further, we compare and analyse existing PDS platforms in terms of their capabilities and core components. Subsequently, we summarise the major challenges and issues facing PDS platforms' development and widespread adoption.
RESUMO
The study of environmental sound classification (ESC) has become popular over the years due to the intricate nature of environmental sounds and the evolution of deep learning (DL) techniques. Forest ESC is one use case of ESC, which has been widely experimented with recently to identify illegal activities inside a forest. However, at present, there is a limitation of public datasets specific to all the possible sounds in a forest environment. Most of the existing experiments have been done using generic environment sound datasets such as ESC-50, U8K, and FSD50K. Importantly, in DL-based sound classification, the lack of quality data can cause misguided information, and the predictions obtained remain questionable. Hence, there is a requirement for a well-defined benchmark forest environment sound dataset. This paper proposes FSC22, which fills the gap of a benchmark dataset for forest environmental sound classification. It includes 2025 sound clips under 27 acoustic classes, which contain possible sounds in a forest environment. We discuss the procedure of dataset preparation and validate it through different baseline sound classification models. Additionally, it provides an analysis of the new dataset compared to other available datasets. Therefore, this dataset can be used by researchers and developers who are working on forest observatory tasks.
Assuntos
Acústica , Som , Benchmarking , Confiabilidade dos Dados , FlorestasRESUMO
The incorporation of deep-learning techniques in embedded systems has enhanced the capabilities of edge computing to a great extent. However, most of these solutions rely on high-end hardware and often require a high processing capacity, which cannot be achieved with resource-constrained edge computing. This study presents a novel approach and a proof of concept for a hardware-efficient automated license plate recognition system for a constrained environment with limited resources. The proposed solution is purely implemented for low-resource edge devices and performed well for extreme illumination changes such as day and nighttime. The generalisability of the proposed models has been achieved using a novel set of neural networks for different hardware configurations based on the computational capabilities and low cost. The accuracy, energy efficiency, communication, and computational latency of the proposed models are validated using different license plate datasets in the daytime and nighttime and in real time. Meanwhile, the results obtained from the proposed study have shown competitive performance to the state-of-the-art server-grade hardware solutions as well.
Assuntos
Computadores , Redes Neurais de ComputaçãoRESUMO
Pharmacological MRI (phMRI) studies seek to capture changes in brain haemodynamics in response to a drug. This provides a methodological platform for the evaluation of novel therapeutics, and when applied to disease states, may provide diagnostic or mechanistic information pertaining to common brain disorders such as dementia. Changes to brain perfusion and blood-cerebrospinal fluid barrier (BCSFB) function can be probed, non-invasively, by arterial spin labelling (ASL) and blood-cerebrospinal fluid barrier arterial spin labelling (BCSFB-ASL) MRI respectively. Here, we introduce a method for simultaneous recording of pharmacological perturbation of brain perfusion and BCSFB function using interleaved echo-time ASL, applied to the anesthetized mouse brain. Using this approach, we capture an exclusive decrease in BCSFB-mediated delivery of arterial blood water to ventricular CSF, following anti-diuretic hormone, vasopressin, administration. The commonly used vasodilatory agent, CO2, induced similar increases (~21%) in both cortical perfusion and the BCSFB-ASL signal. Furthermore, we present evidence that caffeine administration triggers a marked decrease in BCSFB-mediated labelled water delivery (41%), with no significant changes in cortical perfusion. Finally, we demonstrate a marked decrease in the functional response of the BCSFB to, vasopressin, in the aged vs adult brain. Together these data, the first of such kind, highlight the value of this translational approach to capture simultaneous and differential pharmacological modulation of vessel tone at the blood brain barrier and BCSFB and how this relationship may be modified in the ageing brain.
Assuntos
Barreira Hematoencefálica/diagnóstico por imagem , Encéfalo/diagnóstico por imagem , Circulação Cerebrovascular/efeitos dos fármacos , Imageamento por Ressonância Magnética/métodos , Animais , Barreira Hematoencefálica/efeitos dos fármacos , Encéfalo/efeitos dos fármacos , Cafeína/farmacologia , Masculino , Camundongos , Marcadores de Spin , Vasoconstritores/farmacologia , Vasopressinas/farmacologiaRESUMO
Chronic hypertension is a major risk factor for the development of neurodegenerative disease, yet the etiology of hypertension-driven neurodegeneration remains poorly understood. Forming a unique interface between the systemic circulation and the brain, the blood-cerebrospinal fluid barrier (BCSFB) at the choroid plexus (CP) has been proposed as a key site of vulnerability to hypertension that may initiate downstream neurodegenerative processes. However, our ability to understand BCSFB's role in pathological processes has, to date, been restricted by a lack of non-invasive functional measurement techniques. In this work, we apply a novel Blood-Cerebrospinal Fluid Barrier Arterial Spin Labeling (BCSFB-ASL) Magnetic resonance imaging (MRI) approach with the aim of detecting possible derangement of BCSFB function in the Spontaneous Hypertensive Rat (SHR) model using a non-invasive, translational technique. SHRs displayed a 36% reduction in BCSFB-mediated labeled arterial water delivery into ventricular cerebrospinal fluid (CSF), relative to normotensive controls, indicative of down-regulated choroid plexus function. This was concomitant with additional changes in brain fluid biomarkers, namely ventriculomegaly and changes in CSF composition, as measured by T1 lengthening. However, cortical cerebral blood flow (CBF) measurements, an imaging biomarker of cerebrovascular health, revealed no measurable change between the groups. Here, we provide the first demonstration of BCSFB-ASL in the rat brain, enabling non-invasive assessment of BCSFB function in healthy and hypertensive rats. Our data highlights the potential for BCSFB-ASL to serve as a sensitive early biomarker for hypertension-driven neurodegeneration, in addition to investigating the mechanisms relating hypertension to neurodegenerative outcomes.
RESUMO
Doxorubicin (DOX) is a widely used chemotherapeutic agent that can cause serious cardiotoxic side effects culminating in congestive heart failure (HF). There are currently no clinical imaging techniques or biomarkers available to detect DOX-cardiotoxicity before functional decline. Mitochondrial dysfunction is thought to be a key factor driving functional decline, though real-time metabolic fluxes have never been assessed in DOX-cardiotoxicity. Hyperpolarized magnetic resonance imaging (MRI) can assess real-time metabolic fluxes in vivo. Here we show that cardiac functional decline in a clinically relevant rat-model of DOX-HF is preceded by a change in oxidative mitochondrial carbohydrate metabolism, measured by hyperpolarized MRI. The decreased metabolic fluxes were predominantly due to mitochondrial loss and additional mitochondrial dysfunction, and not, as widely assumed hitherto, to oxidative stress. Since hyperpolarized MRI has been successfully translated into clinical trials this opens up the potential to test cancer patients receiving DOX for early signs of cardiotoxicity.