Search

This document discusses the advancements in deep learning technology and image sensing. It highlights the role of deep learning in improving image recognition data and sensor performance in various applications, such as autonomous driving and smartphone camera functionalities. The document presents eleven papers published in a special issue on deep learning technology and image sensing, covering topics like medical imaging, image enhancement, object detection, and innovation in image sensing technologies. The papers demonstrate innovative approaches and methodologies in diverse domains, including liver segmentation, brain tumor classification, optic cup and disc edge detection, image-to-image translation in astronomy, super-resolution imaging, and object detection systems. These studies contribute to the interdisciplinary field of deep learning-based imaging and sensing and offer valuable insights for various fields, including healthcare, astronomy, and robotics. [Extracted from the article]

This paper is on the autonomous detection of humans in off-limits mountains. In off-limits mountains, a human rarely exists, thus human detection is an extremely rare event. Due to the advances in artificial intelligence, object detection–classification algorithms based on a Convolution Neural Network (CNN) can be used for this application. However, considering off-limits mountains, there should be no person in general. Thus, it is not desirable to run object detection–classification algorithms continuously, since they are computationally heavy. This paper addresses a time-efficient human detector system, based on both motion detection and object classification. The proposed scheme is to run a motion detection algorithm from time to time. In the camera image, we define a feasible human space where a human can appear. Once motion is detected inside the feasible human space, one enables the object classification, only inside the bounding box where motion is detected. Since motion detection inside the feasible human space runs much faster than an object detection–classification method, the proposed approach is suitable for real-time human detection with low computational loads. As far as we know, no paper in the literature used the feasible human space, as in our paper. The outperformance of our human detector system is verified by comparing it with other state-of-the-art object detection–classification algorithms (HOG detector, YOLOv7 and YOLOv7-tiny) under experiments. This paper demonstrates that the accuracy of the proposed human detector system is comparable to other state-of-the-art algorithms, while outperforming in computational speed. Our experiments show that in environments with no humans, the proposed human detector runs 62 times faster than YOLOv7 method, while showing comparable accuracy. [ABSTRACT FROM AUTHOR]

This paper addresses issues concerning biometric authentication based on handwritten signatures. Our research aimed to check whether a handwritten signature acquired with a mobile device can effectively verify a user's identity. We present a novel online signature verification method using coordinates of points and pressure values at each point collected with a mobile device. Convolutional neural networks are used for signature verification. In this paper, three neural network models are investigated, i.e., two self-made light SigNet and SigNetExt models and the VGG-16 model commonly used in image processing. The convolutional neural networks aim to determine whether the acquired signature sample matches the class declared by the signer. Thus, the scenario of closed set verification is performed. The effectiveness of our method was tested on signatures acquired with mobile phones. We used the subset of the multimodal database, MobiBits, that was captured using a custom-made application and consists of samples acquired from 53 people of diverse ages. The experimental results on accurate data demonstrate that developed architectures of deep neural networks can be successfully used for online handwritten signature verification. We achieved an equal error rate (EER) of 0.63% for random forgeries and 6.66% for skilled forgeries. [ABSTRACT FROM AUTHOR]

The identification of gas leakages is a significant factor to be taken into consideration in various industries such as coal mines, chemical industries, etc., as well as in residential applications. In order to reduce damage to the environment as well as human lives, early detection and gas type identification are necessary. The main focus of this paper is multimodal gas data that were obtained simultaneously by using multiple sensors for gas detection and a thermal imaging camera. As the reliability and sensitivity of low-cost sensors are less, they are not suitable for gas detection over long distances. In order to overcome the drawbacks of relying just on sensors to identify gases, a thermal camera capable of detecting temperature changes is also used in the collection of the current multimodal dataset The multimodal dataset comprises 6400 samples, including smoke, perfume, a combination of both, and neutral environments. In this paper, convolutional neural networks (CNNs) are trained on thermal image data, utilizing variants such as bidirectional long–short-term memory (Bi-LSTM), dense LSTM, and a fusion of both datasets to effectively classify comma separated value (CSV) data from gas sensors. The dataset can be used as a valuable source for research scholars and system developers to improvise their artificial intelligence (AI) models used for gas leakage detection. Furthermore, in order to ensure the privacy of the client's data, this paper explores the implementation of federated learning for privacy-protected gas leakage classification, demonstrating comparable accuracy to traditional deep learning approaches. [ABSTRACT FROM AUTHOR]

In this paper, artificial intelligence (AI) technology is applied to the electromagnetic imaging of anisotropic objects. Advances in magnetic anomaly sensing systems and electromagnetic imaging use electromagnetic principles to detect and characterize subsurface or hidden objects. We use measured multifrequency scattered fields to calculate the initial dielectric constant distribution of anisotropic objects through the backpropagation scheme (BPS). Later, the estimated multifrequency permittivity distribution is input to a convolutional neural network (CNN) for the adaptive moment estimation (ADAM) method to reconstruct a more accurate image. In the meantime, we also improve the definition of loss function in the CNN. Numerical results show that the improved loss function unifying the structural similarity index measure (SSIM) and root mean square error (RMSE) can effectively enhance image quality. In our simulation environment, noise interference is considered for both TE (transverse electric) and TM (transverse magnetic) waves to reconstruct anisotropic scatterers. Lastly, we conclude that multifrequency reconstructions are more stable and precise than single-frequency reconstructions. [ABSTRACT FROM AUTHOR]

Generative AI has gained enormous interest nowadays due to new applications like ChatGPT, DALL E, Stable Diffusion, and Deepfake. In particular, DALL E, Stable Diffusion, and others (Adobe Firefly, ImagineArt, etc.) can create images from a text prompt and are even able to create photorealistic images. Due to this fact, intense research has been performed to create new image forensics applications able to distinguish between real captured images and videos and artificial ones. Detecting forgeries made with Deepfake is one of the most researched issues. This paper is about another kind of forgery detection. The purpose of this research is to detect photorealistic AI-created images versus real photos coming from a physical camera. Id est, making a binary decision over an image, asking whether it is artificially or naturally created. Artificial images do not need to try to represent any real object, person, or place. For this purpose, techniques that perform a pixel-level feature extraction are used. The first one is Photo Response Non-Uniformity (PRNU). PRNU is a special noise due to imperfections on the camera sensor that is used for source camera identification. The underlying idea is that AI images will have a different PRNU pattern. The second one is error level analysis (ELA). This is another type of feature extraction traditionally used for detecting image editing. ELA is being used nowadays by photographers for the manual detection of AI-created images. Both kinds of features are used to train convolutional neural networks to differentiate between AI images and real photographs. Good results are obtained, achieving accuracy rates of over 95%. Both extraction methods are carefully assessed by computing precision/recall and F1-score measurements. [ABSTRACT FROM AUTHOR]

Partially automated robotic systems, such as camera holders, represent a pivotal step towards enhancing efficiency and precision in surgical procedures. Therefore, this paper introduces an approach for real-time tool localization in laparoscopy surgery using convolutional neural networks. The proposed model, based on two Hourglass modules in series, can localize up to two surgical tools simultaneously. This study utilized three datasets: the ITAP dataset, alongside two publicly available datasets, namely Atlas Dione and EndoVis Challenge. Three variations of the Hourglass-based models were proposed, with the best model achieving high accuracy (92.86%) and frame rates (27.64 FPS), suitable for integration into robotic systems. An evaluation on an independent test set yielded slightly lower accuracy, indicating limited generalizability. The model was further analyzed using the Grad-CAM technique to gain insights into its functionality. Overall, this work presents a promising solution for automating aspects of laparoscopic surgery, potentially enhancing surgical efficiency by reducing the need for manual endoscope manipulation. [ABSTRACT FROM AUTHOR]

The latest survey results show an increase in accidents on the roads involving pedestrians and cyclists. The reasons for such situations are many, the fault actually lies on both sides. Equipping vehicles, especially autonomous vehicles, with frequency-modulated continuous-wave (FMCW) radar and dedicated algorithms for analyzing signals in the time–frequency domain as well as algorithms for recognizing objects in radar imaging through deep neural networks can positively affect safety. This paper presents a method for recognizing and distinguishing a group of objects based on radar signatures of objects and a special convolutional neural network structure. The proposed approach is based on a database of radar signatures generated on pedestrian, cyclist, and car models in a Matlab environment. The obtained results of simulations and positive tests provide a basis for the application of the system in many sectors and areas of the economy. Innovative aspects of the work include the method of discriminating between multiple objects on a single radar signature, the dedicated architecture of the convolutional neural network, and the use of a method of generating a custom input database. [ABSTRACT FROM AUTHOR]

Facial emotion recognition (FER) is a computer vision process aimed at detecting and classifying human emotional expressions. FER systems are currently used in a vast range of applications from areas such as education, healthcare, or public safety; therefore, detection and recognition accuracies are very important. Similar to any computer vision task based on image analyses, FER solutions are also suitable for integration with artificial intelligence solutions represented by different neural network varieties, especially deep neural networks that have shown great potential in the last years due to their feature extraction capabilities and computational efficiency over large datasets. In this context, this paper reviews the latest developments in the FER area, with a focus on recent neural network models that implement specific facial image analysis algorithms to detect and recognize facial emotions. This paper's scope is to present from historical and conceptual perspectives the evolution of the neural network architectures that proved significant results in the FER area. This paper endorses convolutional neural network (CNN)-based architectures against other neural network architectures, such as recurrent neural networks or generative adversarial networks, highlighting the key elements and performance of each architecture, and the advantages and limitations of the proposed models in the analyzed papers. Additionally, this paper presents the available datasets that are currently used for emotion recognition from facial expressions and micro-expressions. The usage of FER systems is also highlighted in various domains such as healthcare, education, security, or social IoT. Finally, open issues and future possible developments in the FER area are identified. [ABSTRACT FROM AUTHOR]

Reconfigurable intelligent surfaces (RIS) offer the potential to customize the radio propagation environment for wireless networks, and will be a key element for 6G communications. However, due to the unique constraints in these systems, the optimization problems associated to RIS configuration are challenging to solve. This paper illustrates a new approach to the RIS configuration problem, based on the use of artificial intelligence (AI) and deep learning (DL) algorithms. Concretely, a custom convolutional neural network (CNN) intended for edge computing is presented, and implementations on different representative edge devices are compared, including the use of commercial AI-oriented devices and a field-programmable gate array (FPGA) platform. This FPGA option provides the best performance, with × 20 performance increase over the closest FP32, GPU-accelerated option, and almost × 3 performance advantage when compared with the INT8-quantized, TPU-accelerated implementation. More noticeably, this is achieved even when high-level synthesis (HLS) tools are used and no custom accelerators are developed. At the same time, the inherent reconfigurability of FPGAs opens a new field for their use as enabler hardware in RIS applications. [ABSTRACT FROM AUTHOR]

Imaging is one of the main tools of modern astronomy—many images are collected each day, and they must be processed. Processing such a large amount of images can be complex, time-consuming, and may require advanced tools. One of the techniques that may be employed is artificial intelligence (AI)-based image detection and classification. In this paper, the research is focused on developing such a system for the problem of the Magellan dataset, which contains 134 satellite images of Venus's surface with individual volcanoes marked with circular labels. Volcanoes are classified into four classes depending on their features. In this paper, the authors apply the You-Only-Look-Once (YOLO) algorithm, which is based on a convolutional neural network (CNN). To apply this technique, the original labels are first converted into a suitable YOLO format. Then, due to the relatively small number of images in the dataset, deterministic augmentation techniques are applied. Hyperparameters of the YOLO network are tuned to achieve the best results, which are evaluated as mean average precision (mAP@0.5) for localization accuracy and F1 score for classification accuracy. The experimental results using cross-vallidation indicate that the proposed method achieved 0.835 mAP@0.5 and 0.826 F1 scores, respectively. [ABSTRACT FROM AUTHOR]

The popularity and demand for high-quality date palm fruits (Phoenix dactylifera L.) have been growing, and their quality largely depends on the type of handling, storage, and processing methods. The current methods of geometric evaluation and classification of date palm fruits are characterised by high labour intensity and are usually performed mechanically, which may cause additional damage and reduce the quality and value of the product. Therefore, non-contact methods are being sought based on image analysis, with digital solutions controlling the evaluation and classification processes. The main objective of this paper is to develop an automatic classification model for varieties of date palm fruits using a convolutional neural network (CNN) based on two fundamental criteria, i.e., colour difference and evaluation of geometric parameters of dates. A CNN with a fixed architecture was built, marked as DateNET, consisting of a system of five alternating Conv2D, MaxPooling2D, and Dropout classes. The validation accuracy of the model presented in this study depended on the selection of classification criteria. It was 85.24% for fruit colour-based classification and 87.62% for the geometric parameters only; however, it increased considerably to 93.41% when both the colour and geometry of dates were considered. [ABSTRACT FROM AUTHOR]

The presented paper investigates the problem of endoscopic bleeding detection in endoscopic videos in the form of a binary image classification task. A set of definitions of high-level visual features of endoscopic bleeding is introduced, which incorporates domain knowledge from the field. The high-level features are coupled with respective feature descriptors, enabling automatic capture of the features using image processing methods. Each of the proposed feature descriptors outputs a feature activation map in the form of a grayscale image. Acquired feature maps can be appended in a straightforward way to the original color channels of the input image and passed to the input of a convolutional neural network during the training and inference steps. An experimental evaluation is conducted to compare the classification ROC AUC of feature-extended convolutional neural network models with baseline models using regular color image inputs. The advantage of feature-extended models is demonstrated for the Resnet and VGG convolutional neural network architectures. [ABSTRACT FROM AUTHOR]

In this paper, we propose a compact and low-power mixed-signal approach to implementing convolutional operators that are often responsible for most of the chip area and power consumption of Convolutional Neural Network (CNN) processing chips. The convolutional operators consist of several multiply-and-accumulate (MAC) units. MAC units are the primary components that process convolutional layers and fully connected layers of CNN models. Analog implementation of MAC units opens a new paradigm for realizing low-power CNN processing chips, benefiting from less power and area consumption. The proposed mixed-signal convolutional operator comprises low-power binary-weighted current steering digital-to-analog conversion (DAC) circuits and accumulation capacitors. Compared with a conventional binary-weighted DAC, the proposed circuit benefits from optimum accuracy, smaller area, and lower power consumption due to its symmetric design. The proposed convolutional operator takes as input a set of 9-bit digital input feature data and weight parameters of the convolutional filter. It then calculates the convolutional filter's result and accumulates the resulting voltage on capacitors. In addition, the convolutional operator employs a novel charge-sharing technique to process negative MAC results. We propose an analog max-pooling circuit that instantly selects the maximum input voltage. To demonstrate the performance of the proposed mixed-signal convolutional operator, we implemented a CNN processing chip consisting of 3 analog convolutional operators, with each operator processing a 3 × 3 kernel. This chip contains 27 MAC circuits, an analog max-pooling, and an analog-to-digital conversion (ADC) circuit. The mixed-signal CNN processing chip is implemented using a CMOS 55 nm process, which occupies a silicon area of 0.0559 mm2 and consumes an average power of 540.6 μW. The proposed mixed-signal CNN processing chip offers an area reduction of 84.21% and an energy reduction of 91.85% compared with a conventional digital CNN processing chip. Moreover, another CNN processing chip is implemented with more analog convolutional operators to demonstrate the operation and structure of an example convolutional layer of a CNN model. Therefore, the proposed analog convolutional operator can be adapted in various CNN models as an alternative to digital counterparts. [ABSTRACT FROM AUTHOR]

In this paper, we present the microwave imaging of anisotropic objects by artificial intelligence technology. Since the biaxial anisotropic scatterers have different dielectric constant components in different transverse directions, the problems faced by transverse electronic (TE) polarization waves are more complex than those of transverse magnetic (TM) polarization waves. In other words, measured scattered field information can scarcely reconstruct microwave images due to the high nonlinearity characteristic of TE polarization. Therefore, we first use the dominant current scheme (DCS) and the back-propagation scheme (BPS) to compute the initial guess image. We then apply a trained convolution neural network (CNN) to regenerate the microwave image. Numerical results show that the CNN possesses a good generalization ability under limited training data, which could be favorable to deploy in image processing. Finally, we compare DCS and BPS reconstruction images for anisotropic objects by the CNN and prove that DCS is better than BPS. In brief, successfully reconstructing biaxial anisotropic objects with a CNN is the contribution of this proposal. [ABSTRACT FROM AUTHOR]

Medical images play an important role in medical diagnosis and treatment. Oncologists analyze images to determine the different characteristics of deadly diseases, plan the therapy, and observe the evolution of the disease. The objective of this paper is to propose a method for the detection of brain tumors. Brain tumors are identified from Magnetic Resonance (MR) images by performing suitable segmentation procedures. The latest technical literature concerning radiographic images of the brain shows that deep learning methods can be implemented to extract specific features of brain tumors, aiding clinical diagnosis. For this reason, most data scientists and AI researchers work on Machine Learning methods for designing automatic screening procedures. Indeed, an automated method would result in quicker segmentation findings, providing a robust output with respect to possible differences in data sources, mostly due to different procedures in data recording and storing, resulting in a more consistent identification of brain tumors. To improve the performance of the segmentation procedure, new architectures are proposed and tested in this paper. We propose deep neural networks for the detection of brain tumors, trained on the MRI scans of patients' brains. The proposed architectures are based on convolutional neural networks and inception modules for brain tumor segmentation. A comparison of these proposed architectures with the baseline reference ones shows very interesting results. MI-Unet showed a performance increase in comparison to baseline Unet architecture by 7.5 % in dice score, 23.91 % insensitivity, and 7.09 % in specificity. Depth-wise separable MI-Unet showed a performance increase by 10.83 % in dice score, 2.97 % in sensitivity, and 12.72 % in specificity as compared to the baseline Unet architecture. Hybrid Unet architecture achieved performance improvement of 9.71 % in dice score, 3.56 % in sensitivity, and 12.6 % in specificity. Whereas the depth-wise separable hybrid Unet architecture outperformed the baseline architecture by 15.45 % in dice score, 20.56 % in sensitivity, and 12.22 % in specificity. [ABSTRACT FROM AUTHOR]

Due to the recent increasing utilization of deep learning models on edge devices, the industry demand for Deep Learning Model Optimization (DLMO) is also increasing. This paper derives a usage strategy of DLMO based on the performance evaluation through light convolution, quantization, pruning techniques and knowledge distillation, known to be excellent in reducing memory size and operation delay with a minimal accuracy drop. Through experiments regarding image classification, we derive possible and optimal strategies to apply deep learning into Internet of Things (IoT) or tiny embedded devices. In particular, strategies for DLMO technology most suitable for each on-device Artificial Intelligence (AI) service are proposed in terms of performance factors. In this paper, we suggest a possible solution of the most rational algorithm under very limited resource environments by utilizing mature deep learning methodologies. [ABSTRACT FROM AUTHOR]

Catastrophic forgetting, which means a rapid forgetting of learned representations while learning new data/samples, is one of the main problems of deep neural networks. In this paper, we propose a novel incremental learning framework that can address the forgetting problem by learning new incoming data in an online manner. We develop a new incremental learning framework that can learn extra data or new classes with less catastrophic forgetting. We adopt the hippocampal memory process to the deep neural networks by defining the effective maximum of neural activation and its boundary to represent a feature distribution. In addition, we incorporate incremental QR factorization into the deep neural networks to learn new data with both existing labels and new labels with less forgetting. The QR factorization can provide the accurate subspace prior, and incremental QR factorization can reasonably express the collaboration between new data with both existing classes and new class with less forgetting. In our framework, a set of appropriate features (i.e., nodes) provides improved representation for each class. We apply our method to the convolutional neural network (CNN) for learning Cifar-100 and Cifar-10 datasets. The experimental results show that the proposed method efficiently alleviates the stability and plasticity dilemma in the deep neural networks by providing the performance stability of a trained network while effectively learning unseen data and additional new classes. [ABSTRACT FROM AUTHOR]

A global health emergency resulted from the COVID-19 epidemic. Image recognition techniques are a useful tool for limiting the spread of the pandemic; indeed, the World Health Organization (WHO) recommends the use of face masks in public places as a form of protection against contagion. Hence, innovative systems and algorithms were deployed to rapidly screen a large number of people with faces covered by masks. In this article, we analyze the current state of research and future directions in algorithms and systems for masked-face recognition. First, the paper discusses the importance and applications of facial and face mask recognition, introducing the main approaches. Afterward, we review the recent facial recognition frameworks and systems based on Convolution Neural Networks, deep learning, machine learning, and MobilNet techniques. In detail, we analyze and critically discuss recent scientific works and systems which employ machine learning (ML) and deep learning tools for promptly recognizing masked faces. Also, Internet of Things (IoT)-based sensors, implementing ML and DL algorithms, were described to keep track of the number of persons donning face masks and notify the proper authorities. Afterward, the main challenges and open issues that should be solved in future studies and systems are discussed. Finally, comparative analysis and discussion are reported, providing useful insights for outlining the next generation of face recognition systems. [ABSTRACT FROM AUTHOR]

With a view of the post-COVID-19 world and probable future pandemics, this paper presents an Internet of Things (IoT)-based automated healthcare diagnosis model that employs a mixed approach using data augmentation, transfer learning, and deep learning techniques and does not require physical interaction between the patient and physician. Through a user-friendly graphic user interface and availability of suitable computing power on smart devices, the embedded artificial intelligence allows the proposed model to be effectively used by a layperson without the need for a dental expert by indicating any issues with the teeth and subsequent treatment options. The proposed method involves multiple processes, including data acquisition using IoT devices, data preprocessing, deep learning-based feature extraction, and classification through an unsupervised neural network. The dataset contains multiple periapical X-rays of five different types of lesions obtained through an IoT device mounted within the mouth guard. A pretrained AlexNet, a fast GPU implementation of a convolutional neural network (CNN), is fine-tuned using data augmentation and transfer learning and employed to extract the suitable feature set. The data augmentation avoids overtraining, whereas accuracy is improved by transfer learning. Later, support vector machine (SVM) and the K-nearest neighbors (KNN) classifiers are trained for lesion classification. It was found that the proposed automated model based on the AlexNet extraction mechanism followed by the SVM classifier achieved an accuracy of 98%, showing the effectiveness of the presented approach. [ABSTRACT FROM AUTHOR]

Since the 20th century, a rapid process of motorization has begun. The main goal of researchers, engineers and technology companies is to increase the safety and optimality of the movement of vehicles, as well as to reduce the environmental damage caused by the automotive industry. The difficulty of managing traffic flows is that cars are driven by a person and their behavior, even in similar situations, is different and difficult to predict. To solve this problem, ground-based unmanned vehicles are increasingly being developed and implemented; however, like any other intelligent system, it is necessary to train different road scenarios. Currently, an engineer is driving an unmanned vehicle for training and thousands of kilometers are being driven for training. Of course, this approach to training unmanned vehicles is very long, and it is impossible to reproduce all the scenarios that can be found in real operations on a real road. Based on this, we offer a simulator of a realistic urban environment which allows you to reduce the training time and allows you to generate all kinds of events. To implement such a simulator, it is necessary to develop a method that would allow recreating a realistic world in one passage with cameras (monocular) installed on board the vehicle. Based on this, the purpose of this work is to develop an intelligent vehicle recognition system using convolutional neural networks, which allows you to create mesh objects for further placement in the simulator. It is important to note that the resulting objects should be optimal in size so as not to overload the system, since a large number of road infrastructure objects are stored there. Also, neural complexity should not be excessive. In this paper, the general concept and classification of convolutional neural networks are given, which allow solving the problem of recognizing 3D objects in images. Based on the analysis, the existing neural network architectures do not solve the problems mentioned above. In this connection, the authors first of all carried out the design of the system according to the methodology of modeling business processes, and also modified and developed the architecture of the neural network, which allows classifying objects with sufficient accuracy, obtaining optimized mesh objects and reducing computational complexity. The methods proposed in this paper are used in a simulator of a realistic urban environment, which reduces the time and computational costs when training unmanned transport systems. [ABSTRACT FROM AUTHOR]

Traffic Sign Recognition (TSR) is one of the many utilities made possible by embedded systems with internet connections. Through the usage of vehicular cameras, it's possible to capture and classify traffic signs in real time with Artificial Intelligence (AI), more specifically, Convolutional Neural Networks (CNNs) based techniques. This article discusses the implementation of such TSR systems, and the building process of datasets for AI training. Such datasets include a brand new class to be used in TSR, vegetation occlusion. The results show that this approach is useful in making traffic sign maintenance faster since this application turns vehicles into moving sensors in that context. Leaning on the proposed technique, identified irregularities in traffic signs can be reported to a responsible body so they will eventually be fixed, contributing to a safer traffic environment. This paper also discusses the usage and performance of different YOLO models according to our case studies. [ABSTRACT FROM AUTHOR]

Over the past few years, several applications have been extensively exploiting the advantages of deep learning, in particular when using convolutional neural networks (CNNs). The intrinsic flexibility of such models makes them widely adopted in a variety of practical applications, from medical to industrial. In this latter scenario, however, using consumer Personal Computer (PC) hardware is not always suitable for the potential harsh conditions of the working environment and the strict timing that industrial applications typically have. Therefore, the design of custom FPGA (Field Programmable Gate Array) solutions for network inference is gaining massive attention from researchers and companies as well. In this paper, we propose a family of network architectures composed of three kinds of custom layers working with integer arithmetic with a customizable precision (down to just two bits). Such layers are designed to be effectively trained on classical GPUs (Graphics Processing Units) and then synthesized to FPGA hardware for real-time inference. The idea is to provide a trainable quantization layer, called Requantizer, acting both as a non-linear activation for neurons and a value rescaler to match the desired bit precision. This way, the training is not only quantization-aware, but also capable of estimating the optimal scaling coefficients to accommodate both the non-linear nature of the activations and the constraints imposed by the limited precision. In the experimental section, we test the performance of this kind of model while working both on classical PC hardware and a case-study implementation of a signal peak detection device running on a real FPGA. We employ TensorFlow Lite for training and comparison, and use Xilinx FPGAs and Vivado for synthesis and implementation. The results show an accuracy of the quantized networks close to the floating point version, without the need for representative data for calibration as in other approaches, and performance that is better than dedicated peak detection algorithms. The FPGA implementation is able to run in real time at a rate of four gigapixels per second with moderate hardware resources, while achieving a sustained efficiency of 0.5 TOPS/W (tera operations per second per watt), in line with custom integrated hardware accelerators. [ABSTRACT FROM AUTHOR]

Developing computer-aided approaches for cancer diagnosis and grading is currently receiving an increasing demand: this could take over intra- and inter-observer inconsistency, speed up the screening process, increase early diagnosis, and improve the accuracy and consistency of the treatment-planning processes.The third most common cancer worldwide and the second most common in women is colorectal cancer (CRC). Grading CRC is a key task in planning appropriate treatments and estimating the response to them. Unfortunately, it has not yet been fully demonstrated how the most advanced models and methodologies of machine learning can impact this crucial task.This paper systematically investigates the use of advanced deep models (convolutional neural networks and transformer architectures) to improve colon carcinoma detection and grading from histological images. To the best of our knowledge, this is the first attempt at using transformer architectures and ensemble strategies for exploiting deep learning paradigms for automatic colon cancer diagnosis. Results on the largest publicly available dataset demonstrated a substantial improvement with respect to the leading state-of-the-art methods. In particular, by exploiting a transformer architecture, it was possible to observe a 3% increase in accuracy in the detection task (two-class problem) and up to a 4% improvement in the grading task (three-class problem) by also integrating an ensemble strategy. [ABSTRACT FROM AUTHOR]

Human Activity Recognition (HAR) is a complex problem in deep learning, and One-Dimensional Convolutional Neural Networks (1D CNNs) have emerged as a popular approach for addressing it. These networks efficiently learn features from data that can be utilized to classify human activities with high performance. However, understanding and explaining the features learned by these networks remains a challenge. This paper presents a novel eXplainable Artificial Intelligence (XAI) method for generating visual explanations of features learned by one-dimensional CNNs in its training process, utilizing t-Distributed Stochastic Neighbor Embedding (t-SNE). By applying this method, we provide insights into the decision-making process through visualizing the information obtained from the model's deepest layer before classification. Our results demonstrate that the learned features from one dataset can be applied to differentiate human activities in other datasets. Our trained networks achieved high performance on two public databases, with 0.98 accuracy on the SHO dataset and 0.93 accuracy on the HAPT dataset. The visualization method proposed in this work offers a powerful means to detect bias issues or explain incorrect predictions. This work introduces a new type of XAI application, enhancing the reliability and practicality of CNN models in real-world scenarios. [ABSTRACT FROM AUTHOR]

In this paper, a structural health monitoring (SHM) system is proposed to provide automatic early warning for detecting damage and its location in composite pipelines at an early stage. The study considers a basalt fiber reinforced polymer (BFRP) pipeline with an embedded Fiber Bragg grating (FBG) sensory system and first discusses the shortcomings and challenges with incorporating FBG sensors for accurate detection of damage information in pipelines. The novelty and the main focus of this study is, however, a proposed approach that relies on designing an integrated sensing-diagnostic SHM system that has the capability to detect damage in composite pipelines at an early stage via implementation of an artificial intelligence (AI)-based algorithm combining deep learning and other efficient machine learning methods using an Enhanced Convolutional Neural Network (ECNN) without retraining the model. The proposed architecture replaces the softmax layer by a k-Nearest Neighbor (k-NN) algorithm for inference. Finite element models are developed and calibrated by the results of pipe measurements under damage tests. The models are then used to assess the patterns of the strain distributions of the pipeline under internal pressure loading and under pressure changes due to bursts, and to find the relationship of strains at different locations axially and circumferentially. A prediction algorithm for pipe damage mechanisms using distributed strain patterns is also developed. The ECNN is designed and trained to identify the condition of pipe deterioration so the initiation of damage can be detected. The strain results from the current method and the available experimental results in the literature show excellent agreement. The average error between the ECNN data and FBG sensor data is 0.093%, thus confirming the reliability and accuracy of the proposed method. The proposed ECNN achieves high performance with 93.33 % accuracy (P%), 91.18 % regression rate (R%) and a 90.54 % F1-score (F%). [ABSTRACT FROM AUTHOR]

It is considered that 1 in 10 adults worldwide have diabetes. Diabetic foot ulcers are some of the most common complications of diabetes, and they are associated with a high risk of lower-limb amputation and, as a result, reduced life expectancy. Timely detection and periodic ulcer monitoring can considerably decrease amputation rates. Recent research has demonstrated that computer vision can be used to identify foot ulcers and perform non-contact telemetry by using ulcer and tissue area segmentation. However, the applications are limited to controlled lighting conditions, and expert knowledge is required for dataset annotation. This paper reviews the latest publications on the use of artificial intelligence for ulcer area detection and segmentation. The PRISMA methodology was used to search for and select articles, and the selected articles were reviewed to collect quantitative and qualitative data. Qualitative data were used to describe the methodologies used in individual studies, while quantitative data were used for generalization in terms of dataset preparation and feature extraction. Publicly available datasets were accounted for, and methods for preprocessing, augmentation, and feature extraction were evaluated. It was concluded that public datasets can be used to form a bigger, more diverse datasets, and the prospects of wider image preprocessing and the adoption of augmentation require further research. [ABSTRACT FROM AUTHOR]

With the development of mobile payment, the Internet of Things (IoT) and artificial intelligence (AI), smart vending machines, as a kind of unmanned retail, are moving towards a new future. However, the scarcity of data in vending machine scenarios is not conducive to the development of its unmanned services. This paper focuses on using machine learning on small data to detect the placement of the spiral rack indicated by the end of the spiral rack, which is the most crucial factor in causing a product potentially to get stuck in vending machines during the dispensation. To this end, we propose a k-means clustering-based method for splitting small data that is unevenly distributed both in number and in features due to real-world constraints and design a remarkably lightweight convolutional neural network (CNN) as a classifier model for the benefit of real-time application. Our proposal of data splitting along with the CNN is visually interpreted to be effective in that the trained model is robust enough to be unaffected by changes in products and reaches an accuracy of 100 % . We also design a single-board computer-based handheld device and implement the trained model to demonstrate the feasibility of a real-time application. [ABSTRACT FROM AUTHOR]

Speech recognition has progressed tremendously in the area of artificial intelligence (AI). However, the performance of the real-time offline Chinese speech recognition neural network accelerator for edge AI needs to be improved. This paper proposes a configurable convolutional neural network accelerator based on a lightweight speech recognition model, which can dramatically reduce hardware resource consumption while guaranteeing an acceptable error rate. For convolutional layers, the weights are binarized to reduce the number of model parameters and improve computational and storage efficiency. A multichannel shared computation (MCSC) architecture is proposed to maximize the reuse of weight and feature map data. The binary weight-sharing processing engine (PE) is designed to avoid limiting the number of multipliers. A custom instruction set is established according to the variable length of voice input to configure parameters for adapting to different network structures. Finally, the ping-pong storage method is used when the feature map is an input. We implemented this accelerator on Xilinx ZYNQ XC7Z035 under the working frequency of 150 MHz. The processing time for 2.24 s and 8 s of speech was 69.8 ms and 189.51 ms, respectively, and the convolution performance reached 35.66 GOPS/W. Compared with other computing platforms, accelerators perform better in terms of energy efficiency, power consumption and hardware resource consumption. [ABSTRACT FROM AUTHOR]

The amount of road accidents caused by driver drowsiness is one of the world's major challenges. These accidents lead to numerous fatal and non-fatal injuries which impose substantial financial strain on individuals and governments every year. As a result, it is critical to prevent catastrophic accidents and reduce the financial burden on society caused by driver drowsiness. The research community has primarily focused on two approaches to identify driver drowsiness during the last decade: intrusive and non-intrusive. The intrusive approach includes physiological measures, and the non-intrusive approach includes vehicle-based and behavioral measures. In an intrusive approach, sensors are used to detect driver drowsiness by placing them on the driver's body, whereas in a non-intrusive approach, a camera is used for drowsiness detection by identifying yawning patterns, eyelid movement and head inclination. Noticeably, most research has been conducted in driver drowsiness detection methods using only single measures that failed to produce good outcomes. Furthermore, these measures were only functional in certain conditions. This paper proposes a model that combines the two approaches, non-intrusive and intrusive, to detect driver drowsiness. Behavioral measures as a non-intrusive approach and sensor-based physiological measures as an intrusive approach are combined to detect driver drowsiness. The proposed hybrid model uses AI-based Multi-Task Cascaded Convolutional Neural Networks (MTCNN) as a behavioral measure to recognize the driver's facial features, and the Galvanic Skin Response (GSR) sensor as a physiological measure to collect the skin conductance of the driver that helps to increase the overall accuracy. Furthermore, the model's efficacy has been computed in a simulated environment. The outcome shows that the proposed hybrid model is capable of identifying the transition from awake to a drowsy state in the driver in all conditions with the efficacy of 91%. [ABSTRACT FROM AUTHOR]

Tele-rehabilitation has the potential to considerably change the way patients are monitored from their homes during the care process, by providing equitable access without the need to travel to rehab centers or shoulder the high cost of personal in-home services. Developing a tele-rehab platform with the capability of automating exercise guidance is likely to have a significant impact on rehabilitation outcomes. In this paper, a new vision-based biofeedback system is designed and validated to identify the quality of performed exercises. This new system will help patients to refine their movements to get the most out of their plan of care. An open dataset was used, which consisted of data from 30 participants performing nine different exercises. Each exercise was labeled as "Correctly" or "Incorrectly" executed by five clinicians. We used a pre-trained 3D Convolution Neural Network (3D-CNN) to design our biofeedback system. The proposed system achieved average accuracy values of 90.57% ± 9.17% and 83.78% ± 7.63% using 10-Fold and Leave-One-Subject-Out (LOSO) cross validation, respectively. In addition, we obtained average F1-scores of 71.78% ± 5.68% using 10-Fold and 60.64% ± 21.3% using LOSO validation. The proposed 3D-CNN was able to classify the rehabilitation videos and feedback on the quality of exercises to help users modify their movement patterns. [ABSTRACT FROM AUTHOR]

This paper describes the process of developing a classification model for the effective detection of malignant melanoma, an aggressive type of cancer in skin lesions. Primary focus is given on fine-tuning and improving a state-of-the-art convolutional neural network (CNN) to obtain the optimal ROC-AUC score. The study investigates a variety of artificial intelligence (AI) clustering techniques to train the developed models on a combined dataset of images across data from the 2019 and 2020 IIM-ISIC Melanoma Classification Challenges. The models were evaluated using varying cross-fold validations, with the highest ROC-AUC reaching a score of 99.48%. [ABSTRACT FROM AUTHOR]

The acoustic emission (AE) technique is one of the most widely used in the field of structural monitoring. Its popularity mainly stems from the fact that it belongs to the category of non-destructive techniques (NDT) and allows the passive monitoring of structures. The technique employs piezoelectric sensors to measure the elastic ultrasonic wave that propagates in the material as a result of the crack formation's abrupt release of energy. The recorded signal can be investigated to obtain information about the source crack, its position, and its typology (Mode I, Mode II). Over the years, many techniques have been developed for the localization, characterization, and quantification of damage from the study of acoustic emission. The onset time of the signal is an essential information item to be derived from waveform analysis. This information combined with the use of the triangulation technique allows for the identification of the crack location. In the literature, it is possible to find many methods to identify, with increasing accuracy, the onset time of the P-wave. Indeed, the precision of the onset time detection affects the accuracy of identifying the location of the crack. In this paper, two techniques for the definition of the onset time of acoustic emission signals are presented. The first method is based on the Akaike Information Criterion (AIC) while the second one relies on the use of artificial intelligence (AI). A recurrent convolutional neural network (R-CNN) designed for sound event detection (SED) is trained on three different datasets composed of seismic signals and acoustic emission signals to be tested on a real-world acoustic emission dataset. The new method allows taking advantage of the similarities between acoustic emissions, seismic signals, and sound signals, enhancing the accuracy in determining the onset time. [ABSTRACT FROM AUTHOR]

The novel coronavirus (COVID-19), which emerged as a pandemic, has engulfed so many lives and affected millions of people across the world since December 2019. Although this disease is under control nowadays, yet it is still affecting people in many countries. The traditional way of diagnosis is time taking, less efficient, and has a low rate of detection of this disease. Therefore, there is a need for an automatic system that expedites the diagnosis process while retaining its performance and accuracy. Artificial intelligence (AI) technologies such as machine learning (ML) and deep learning (DL) potentially provide powerful solutions to address this problem. In this study, a state-of-the-art CNN model densely connected squeeze convolutional neural network (DCSCNN) has been developed for the classification of X-ray images of COVID-19, pneumonia, normal, and lung opacity patients. Data were collected from different sources. We applied different preprocessing techniques to enhance the quality of images so that our model could learn accurately and give optimal performance. Moreover, the attention regions and decisions of the AI model were visualized using the Grad-CAM and LIME methods. The DCSCNN combines the strength of the Dense and Squeeze networks. In our experiment, seven kinds of classification have been performed, in which six are binary classifications (COVID vs. normal, COVID vs. lung opacity, lung opacity vs. normal, COVID vs. pneumonia, pneumonia vs. lung opacity, pneumonia vs. normal) and one is multiclass classification (COVID vs. pneumonia vs. lung opacity vs. normal). The main contributions of this paper are as follows. First, the development of the DCSNN model which is capable of performing binary classification as well as multiclass classification with excellent classification accuracy. Second, to ensure trust, transparency, and explainability of the model, we applied two popular Explainable AI techniques (XAI). i.e., Grad-CAM and LIME. These techniques helped to address the black-box nature of the model while improving the trust, transparency, and explainability of the model. Our proposed DCSCNN model achieved an accuracy of 98.8% for the classification of COVID-19 vs normal, followed by COVID-19 vs. lung opacity: 98.2%, lung opacity vs. normal: 97.2%, COVID-19 vs. pneumonia: 96.4%, pneumonia vs. lung opacity: 95.8%, pneumonia vs. normal: 97.4%, and lastly for multiclass classification of all the four classes i.e., COVID vs. pneumonia vs. lung opacity vs. normal: 94.7%, respectively. The DCSCNN model provides excellent classification performance consequently, helping doctors to diagnose diseases quickly and efficiently. [ABSTRACT FROM AUTHOR]

The digital optical fiber repeater (DOFR) is an important infrastructure in the LTE networks, which solve the problem of poor regional signal quality. Various types of conventional measurement data from the LTE network cannot indicate whether a working DOFR is present in the cell. Currently, the detection of DOFRs relies solely on maintenance engineers for field detection. Manual detection methods are not timely or efficient, because of the large number and wide geographical distribution of DOFRs. Implementing automatic detection of DOFR can reduce the maintenance cost for mobile network operators. We treat the DOFR detection problem as a classification problem and employ a deep convolutional neural network (DCNN) to tackle it. The measurement report (MR) we used in this paper are tabular data, which is not an ideal input for DCNN. We propose a novel MR representation method that takes the overall MR data of a cell as a sample rather than a single record in the table, and represents the MR data as a pseudo-image matrix (PIM). The PIM will be used as the input for training DCNN, and the trained DCNN will be used to perform DOFR detection tasks. We conducted a series of experiments on real MR data, and the classification accuracy can achieve 93%. The proposed AI-based method can effectively detect the DOFR in a cell. [ABSTRACT FROM AUTHOR]

Electrical impedance tomography (EIT) has been applied in the field of human-computer interaction due to its advantages including the fact that it is non-invasive and has both low power consumption and a low cost. Previous work has focused on static gesture recognition based on EIT. Compared with static gestures, dynamic gestures are more informative and can achieve more functions in human-machine collaboration. In order to verify the feasibility of dynamic gesture recognition based on EIT, a traditional excitation drive pattern is optimized in this paper. The drive pattern of the fixed excitation electrode is tested for the first time to simplify the measurement process of the dynamic gesture. To improve the recognition accuracy of the dynamic gestures, a dual-channel feature extraction network combining a convolutional neural network (CNN) and gated recurrent unit (GRU), namely CG-SVM, is proposed. The new center distance loss is designed in order to simultaneously supervise the intra-class distance and inter-class distance. As a result, the discriminability of the confusing data is improved. With the new excitation drive pattern and classification network, the recognition accuracy of different interference data has increased by 2.7~14.2%. The new method has stronger robustness, and realizes the dynamic gesture recognition based on EIT for the first time. [ABSTRACT FROM AUTHOR]

This paper proposes a development of automatic rib sequence labeling systems on chest computed tomography (CT) images with two suggested methods and three-dimensional (3D) region growing. In clinical practice, radiologists usually define anatomical terms of location depending on the rib's number. Thus, with the manual process of labeling 12 pairs of ribs and counting their sequence, it is necessary to refer to the annotations every time the radiologists read chest CT. However, the process is tedious, repetitive, and time-consuming as the demand for chest CT-based medical readings has increased. To handle the task efficiently, we proposed an automatic rib sequence labeling system and implemented comparison analysis on two methods. With 50 collected chest CT images, we implemented intensity-based image processing (IIP) and a convolutional neural network (CNN) for rib segmentation on this system. Additionally, three-dimensional (3D) region growing was used to classify each rib's label and put in a sequence label. The IIP-based method reported a 92.0% and the CNN-based method reported a 98.0% success rate, which is the rate of labeling appropriate rib sequences over whole pairs (1st to 12th) for all slices. We hope for the applicability thereof in clinical diagnostic environments by this method-efficient automatic rib sequence labeling system. [ABSTRACT FROM AUTHOR]

Convolution Neural Networks (CNNs) are gaining ground in deep learning and Artificial Intelligence (AI) domains, and they can benefit from rapid prototyping in order to produce efficient and low-power hardware designs. The inference process of a Deep Neural Network (DNN) is considered a computationally intensive process that requires hardware accelerators to operate in real-world scenarios due to the low latency requirements of real-time applications. As a result, High-Level Synthesis (HLS) tools are gaining popularity since they provide attractive ways to reduce design time complexity directly in register transfer level (RTL). In this paper, we implement a MobileNetV2 model using a state-of-the-art HLS tool in order to conduct a design space exploration and to provide insights on complex hardware designs which are tailored for DNN inference. Our goal is to combine design methodologies with sparsification techniques to produce hardware accelerators that achieve comparable error metrics within the same order of magnitude with the corresponding state-of-the-art systems while also significantly reducing the inference latency and resource utilization. Toward this end, we apply sparse matrix techniques on a MobileNetV2 model for efficient data representation, and we evaluate our designs in two different weight pruning approaches. Experimental results are evaluated with respect to the CIFAR-10 data set using several different design methodologies in order to fully explore their effects on the performance of the model under examination. [ABSTRACT FROM AUTHOR]

Detecting the objects surrounding a moving vehicle is essential for autonomous driving and for any kind of advanced driving assistance system; such a system can also be used for analyzing the surrounding traffic as the vehicle moves. The most popular techniques for object detection are based on image processing; in recent years, they have become increasingly focused on artificial intelligence. Systems using monocular vision are increasingly popular for driving assistance, as they do not require complex calibration and setup. The lack of three-dimensional data is compensated for by the efficient and accurate classification of the input image pixels. The detected objects are usually identified as cuboids in the 3D space, or as rectangles in the image space. Recently, instance segmentation techniques have been developed that are able to identify the freeform set of pixels that form an individual object, using complex convolutional neural networks (CNNs). This paper presents an alternative to these instance segmentation networks, combining much simpler semantic segmentation networks with light, geometrical post-processing techniques, to achieve instance segmentation results. The semantic segmentation network produces four semantic labels that identify the quarters of the individual objects: top left, top right, bottom left, and bottom right. These pixels are grouped into connected regions, based on their proximity and their position with respect to the whole object. Each quarter is used to generate a complete object hypothesis, which is then scored according to object pixel fitness. The individual homogeneous regions extracted from the labeled pixels are then assigned to the best-fitted rectangles, leading to complete and freeform identification of the pixels of individual objects. The accuracy is similar to instance segmentation-based methods but with reduced complexity in terms of trainable parameters, which leads to a reduced demand for computational resources. [ABSTRACT FROM AUTHOR]

The development of a Social Intelligence System based on artificial intelligence is one of the cutting edge technologies in Assistive Robotics. Such systems need to create an empathic interaction with the users; therefore, it os required to include an Emotion Recognition (ER) framework which has to run, in near real-time, together with several other intelligent services. Most of the low-cost commercial robots, however, although more accessible by users and healthcare facilities, have to balance costs and effectiveness, resulting in under-performing hardware in terms of memory and processing unit. This aspect makes the design of the systems challenging, requiring a trade-off between the accuracy and the complexity of the adopted models. This paper proposes a compact and robust service for Assistive Robotics, called Lightweight EMotion recognitiON (LEMON), which uses image processing, Computer Vision and Deep Learning (DL) algorithms to recognize facial expressions. Specifically, the proposed DL model is based on Residual Convolutional Neural Networks with the combination of Dilated and Standard Convolution Layers. The first remarkable result is the few numbers (i.e., 1.6 Million) of parameters characterizing our model. In addition, Dilated Convolutions expand receptive fields exponentially with preserving resolution, less computation and memory cost to recognize the distinction among facial expressions by capturing the displacement of the pixels. Finally, to reduce the dying ReLU problem and improve the stability of the model, we apply an Exponential Linear Unit (ELU) activation function in the initial layers of the model. We have performed training and evaluation (via one- and five-fold cross validation) of the model with five datasets available in the community and one mixed dataset created by taking samples from all of them. With respect to the other approaches, our model achieves comparable results with a significant reduction in terms of the number of parameters. [ABSTRACT FROM AUTHOR]

Increasing importance in the field of artificial intelligence has led to huge progress in remote sensing. Deep learning approaches have made tremendous progress in hyperspectral image (HSI) classification. However, the complexity in classifying the HSI data using a common convolutional neural network is still a challenge. Further, the network architecture becomes more complex when different spatial–spectral feature information is extracted. Usually, CNN has a large number of trainable parameters, which increases the computational complexity of HSI data. In this paper, an optimized squeeze–excitation AdaBound dense network (SE-AB-DenseNet) is designed to emphasize the significant spatial–spectral features of HSI data. The dense network is combined with the AdaBound and squeeze–excitation modules to give lower computation costs and better classification performance. The AdaBound optimizer gives the proposed model the ability to improve its stability and enhance its classification accuracy by approximately 2%. Additionally, the cutout regularization technique is used for HSI spatial–spectral classification to overcome the problem of overfitting. The experiments were carried out on two commonly used hyperspectral datasets (Indian Pines and Salinas). The experiment results on the datasets show a competitive classification accuracy when compared with state-of-the-art methods with limited training samples. From the SE-AB-DenseNet with the cutout model, the overall accuracies for the Indian Pines and Salinas datasets were observed to be 99.37 and 99.78, respectively. [ABSTRACT FROM AUTHOR]

In the context in which it was demonstrated that humanoid robots are efficient in helping children diagnosed with autism in exploring their affective state, this paper underlines and proves the efficiency of a previously developed machine learning-based mobile application called PandaSays, which was improved and integrated with an Alpha 1 Pro robot, and discusses performance evaluations using deep convolutional neural networks and residual neural networks. The model trained with MobileNet convolutional neural network had an accuracy of 56.25%, performing better than ResNet50 and VGG16. A strategy for commanding the Alpha 1 Pro robot without its native application was also established and a robot module was developed that includes the communication protocols with the application PandaSays. The output of the machine learning algorithm involved in PandaSays is sent to the humanoid robot to execute some actions as singing, dancing, and so on. Alpha 1 Pro has its own programming language—Blockly—and, in order to give the robot specific commands, Bluetooth programming is used, with the help of a Raspberry Pi. Therefore, the robot motions can be controlled based on the corresponding protocols. The tests have proved the robustness of the whole solution. [ABSTRACT FROM AUTHOR]

With the continuous promotion of "smart cities" worldwide, the approach to be used in combining smart cities with modern advanced technologies (Internet of Things, cloud computing, artificial intelligence) has become a hot topic. However, due to the non-stationary nature of environmental sound and the interference of urban noise, it is challenging to fully extract features from the model with a single input and achieve ideal classification results, even with deep learning methods. To improve the recognition accuracy of ESC (environmental sound classification), we propose a dual-branch residual network (dual-resnet) based on feature fusion. Furthermore, in terms of data pre-processing, a loop-padding method is proposed to patch shorter data, enabling it to obtain more useful information. At the same time, in order to prevent the occurrence of overfitting, we use the time-frequency data enhancement method to expand the dataset. After uniform pre-processing of all the original audio, the dual-branch residual network automatically extracts the frequency domain features of the log-Mel spectrogram and log-spectrogram. Then, the two different audio features are fused to make the representation of the audio features more comprehensive. The experimental results show that compared with other models, the classification accuracy of the UrbanSound8k dataset has been improved to different degrees. [ABSTRACT FROM AUTHOR]

The use of artificial intelligence algorithms (AI) has gained importance for dental applications in recent years. Analyzing AI information from different sensor data such as images or panoramic radiographs (panoramic X-rays) can help to improve medical decisions and achieve early diagnosis of different dental pathologies. In particular, the use of deep learning (DL) techniques based on convolutional neural networks (CNNs) has obtained promising results in dental applications based on images, in which approaches based on classification, detection, and segmentation are being studied with growing interest. However, there are still several challenges to be tackled, such as the data quality and quantity, the variability among categories, and the analysis of the possible bias and variance associated with each dataset distribution. This study aims to compare the performance of three deep learning object detection models—Faster R-CNN, YOLO V2, and SSD—using different ResNet architectures (ResNet-18, ResNet-50, and ResNet-101) as feature extractors for detecting and classifying third molar angles in panoramic X-rays according to Winter's classification criterion. Each object detection architecture was trained, calibrated, validated, and tested with three different feature extraction CNNs which are ResNet-18, ResNet-50, and ResNet-101, which were the networks that best fit our dataset distribution. Based on such detection networks, we detect four different categories of angles in third molars using panoramic X-rays by using Winter's classification criterion. This criterion characterizes the third molar's position relative to the second molar's longitudinal axis. The detected categories for the third molars are distoangular, vertical, mesioangular, and horizontal. For training, we used a total of 644 panoramic X-rays. The results obtained in the testing dataset reached up to 99% mean average accuracy performance, demonstrating the YOLOV2 obtained higher effectiveness in solving the third molar angle detection problem. These results demonstrate that the use of CNNs for object detection in panoramic radiographs represents a promising solution in dental applications. [ABSTRACT FROM AUTHOR]

In modern cyber-physical systems, the integration of AI into vision pipelines is now a standard practice for applications ranging from autonomous vehicles to mobile devices. Traditional AI integration often relies on cloud-based processing, which faces challenges such as data access bottlenecks, increased latency, and high power consumption. This article reviews embedded AI vision systems, examining the diverse landscape of near-sensor and in-sensor processing architectures that incorporate convolutional neural networks. We begin with a comprehensive analysis of the critical characteristics and metrics that define the performance of AI-integrated vision systems. These include sensor resolution, frame rate, data bandwidth, computational throughput, latency, power efficiency, and overall system scalability. Understanding these metrics provides a foundation for evaluating how different embedded processing architectures impact the entire vision pipeline, from image capture to AI inference. Our analysis delves into near-sensor systems that leverage dedicated hardware accelerators and commercially available components to efficiently process data close to their source, minimizing data transfer overhead and latency. These systems offer a balance between flexibility and performance, allowing for real-time processing in constrained environments. In addition, we explore in-sensor processing solutions that integrate computational capabilities directly into the sensor. This approach addresses the rigorous demand constraints of embedded applications by significantly reducing data movement and power consumption while also enabling in-sensor feature extraction, pre-processing, and CNN inference. By comparing these approaches, we identify trade-offs related to flexibility, power consumption, and computational performance. Ultimately, this article provides insights into the evolving landscape of embedded AI vision systems and suggests new research directions for the development of next-generation machine vision systems. [ABSTRACT FROM AUTHOR]

Low-Earth-orbit (LEO) satellites are widely acknowledged as a promising infrastructure solution for global Internet of Things (IoT) services. However, the Doppler effect presents a significant challenge in the context of long-range (LoRa) modulation uplink connectivity. This study comprehensively examines the operational efficiency of LEO satellites concerning the Doppler weather effect, with state-of-the-art artificial intelligence techniques. Two LEO satellite constellations—Globalstar and the International Space Station (ISS)—were detected and tracked using ground radars in Perth and Brisbane, Australia, for 24 h starting 1 January 2024. The study involves modelling the constellation, calculating latency, and frequency offset and designing a hybrid Iterative Input Selection–Long Short-Term Memory Network (IIS-LSTM) integrated model to predict the Doppler weather profile for LEO satellites. The IIS algorithm selects relevant input variables for the model, while the LSTM algorithm learns and predicts patterns. This model is compared with Convolutional Neural Network and Extreme Gradient Boosting (XGBoost) models. The results show that the packet delivery rate is above 91% for the sensitive spread factor 12 with a bandwidth of 11.5 MHz for Globalstar and 145.8 MHz for ISS NAUKA. The carrier frequency for ISS orbiting at 402.3 km is 631 MHz and 500 MHz for Globalstar at 1414 km altitude, aiding in combating packet losses. The ISS-LSTM model achieved an accuracy of 97.51% and a loss of 1.17% with signal-to-noise ratios (SNRs) ranging from 0–30 dB. The XGB model has the fastest testing time, attaining ≈0.0997 s for higher SNRs and an accuracy of 87%. However, in lower SNR, it proves to be computationally expensive. IIS-LSTM attains a better computation time for lower SNRs at ≈0.4651 s, followed by XGB at ≈0.5990 and CNN at ≈0.6120 s. The study calls for further research on LoRa Doppler analysis, considering atmospheric attenuation, and relevant space parameters for future work. [ABSTRACT FROM AUTHOR]

Many patients affected by breast cancer die every year because of improper diagnosis and treatment. In recent years, applications of deep learning algorithms in the field of breast cancer detection have proved to be quite efficient. However, the application of such techniques has a lot of scope for improvement. Major works have been done in this field, however it can be made more efficient by the use of transfer learning to get impressive results. In the proposed approach, Convolutional Neural Network (CNN) is complemented with Transfer Learning for increasing the efficiency and accuracy of early detection of breast cancer for better diagnosis. The thought process involved using a pre-trained model, which already had some weights assigned rather than building the complete model from scratch. This paper mainly focuses on ResNet101 based Transfer Learning Model paired with the ImageNet dataset. The proposed framework provided us with an accuracy of 99.58%. Extensive experiments and tuning of hyperparameters have been performed to acquire the best possible results in terms of classification. The proposed frameworks aims to be an efficient tool for all doctors and society as a whole and help the user in early detection of breast cancer. [ABSTRACT FROM AUTHOR]

Music can generate a positive effect in runners' performance and motivation. However, the practical implementation of music intervention during exercise is mostly absent from the literature. Therefore, this paper designs a playback sequence system for joggers by considering music emotion and physiological signals. This playback sequence is implemented by a music selection module that combines artificial intelligence techniques with physiological data and emotional music. In order to make the system operate for a long time, this paper improves the model and selection music module to achieve lower energy consumption. The proposed model obtains fewer FLOPs and parameters by using logarithm scaled Mel-spectrogram as input features. The accuracy, computational complexity, trainable parameters, and inference time are evaluated on the Bi-modal, 4Q emotion, and Soundtrack datasets. The experimental results show that the proposed model is better than that of Sarkar et al. and achieves competitive performance on Bi-modal (84.91%), 4Q emotion (92.04%), and Soundtrack (87.24%) datasets. More specifically, the proposed model reduces the computational complexity and inference time while maintaining the classification accuracy, compared to other models. Moreover, the size of the proposed model for network training is small, which can be applied to mobiles and other devices with limited computing resources. This study designed the overall playback sequence system by considering the relationship between music emotion and physiological situation during exercise. The playback sequence system can be adopted directly during exercise to improve users' exercise efficiency. [ABSTRACT FROM AUTHOR]

Water features (e.g., water quantity and water quality) are one of the most important environmental factors essential to improving climate-change resilience. Remote sensing (RS) technologies empowered by artificial intelligence (AI) have become one of the most demanded strategies to automating water information extraction and thus intelligent monitoring. In this article, we provide a systematic review of the literature that incorporates artificial intelligence and computer vision methods in the water resources sector with a focus on intelligent water body extraction and water quality detection and monitoring through remote sensing. Based on this review, the main challenges of leveraging AI and RS for intelligent water information extraction are discussed, and research priorities are identified. An interactive web application designed to allow readers to intuitively and dynamically review the relevant literature was also developed. [ABSTRACT FROM AUTHOR]