RESUMEN

In infrared detection scenarios, detecting and recognizing low-contrast and small-sized targets has always been a challenge in the field of computer vision, particularly in complex road traffic environments. Traditional target detection methods usually perform poorly when processing infrared small targets, mainly due to their inability to effectively extract key features and the significant feature loss that occurs during feature transmission. To address these issues, this paper proposes a fast detection and recognition model based on a multi-scale self-attention mechanism, specifically for small road targets in infrared detection scenarios. We first introduce and improve the DyHead structure based on the YOLOv8 algorithm, which employs a multi-head self-attention mechanism to capture target features at various scales and enhance the model's perception of small targets. Additionally, to prevent information loss during the feature transmission process via the FPN structure in traditional YOLO algorithms, this paper introduces and enhances the Gather-and-Distribute Mechanism. By computing dependencies between features using self-attention, it reallocates attention weights in the feature maps to highlight important features and suppress irrelevant information. These improvements significantly enhance the model's capability to detect small targets. Moreover, to further increase detection speed, we pruned the network architecture to reduce computational complexity and parameter count, making the model suitable for real-time processing scenarios. Experiments on our self built infrared road traffic dataset (mainly including two types of targets: vehicles and people) show that compared with the baseline, our method achieves a 3.1% improvement in AP and a 2.5% increase in mAP on the VisDrone2019 dataset, showing significant enhancements in both detection accuracy and processing speed for small targets, with improved robustness and adaptability.

RESUMEN

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.

RESUMEN

Simultaneous Localization and Mapping (SLAM) is a crucial function for most autonomous systems, allowing them to both navigate through and create maps of unfamiliar surroundings. Traditional Visual SLAM, also commonly known as VSLAM, relies on frame-based cameras and structured processing pipelines, which face challenges in dynamic or low-light environments. However, recent advancements in event camera technology and neuromorphic processing offer promising opportunities to overcome these limitations. Event cameras inspired by biological vision systems capture the scenes asynchronously, consuming minimal power but with higher temporal resolution. Neuromorphic processors, which are designed to mimic the parallel processing capabilities of the human brain, offer efficient computation for real-time data processing of event-based data streams. This paper provides a comprehensive overview of recent research efforts in integrating event cameras and neuromorphic processors into VSLAM systems. It discusses the principles behind event cameras and neuromorphic processors, highlighting their advantages over traditional sensing and processing methods. Furthermore, an in-depth survey was conducted on state-of-the-art approaches in event-based SLAM, including feature extraction, motion estimation, and map reconstruction techniques. Additionally, the integration of event cameras with neuromorphic processors, focusing on their synergistic benefits in terms of energy efficiency, robustness, and real-time performance, was explored. The paper also discusses the challenges and open research questions in this emerging field, such as sensor calibration, data fusion, and algorithmic development. Finally, the potential applications and future directions for event-based SLAM systems are outlined, ranging from robotics and autonomous vehicles to augmented reality.

RESUMEN

In the tobacco industry, impurity detection is an important prerequisite for ensuring the quality of tobacco. However, in the actual production process, the complex background environment and the variability of impurity shapes can affect the accuracy of impurity detection by tobacco robots, which leads to a decrease in product quality and an increase in health risks. To address this problem, we propose a new online detection method of tobacco impurities for tobacco robot. Firstly, a BCFormer attention mechanism module is designed to effectively mitigate the interference of irrelevant information in the image and improve the network's ability to identify regions of interest. Secondly, a Dual Feature Aggregation (DFA) module is designed and added to Neck to improve the accuracy of tobacco impurities detection by augmenting the fused feature maps with deep semantic and surface location data. Finally, to address the problem that the traditional loss function cannot accurately reflect the distance between two bounding boxes, this paper proposes an optimized loss function to more accurately assess the quality of the bounding boxes. To evaluate the effectiveness of the algorithm, this paper creates a dataset specifically designed to detect tobacco impurities. Experimental results show that the algorithm performs well in identifying tobacco impurities. Our algorithm improved the mAP value by about 3.01% compared to the traditional YOLOX method. The real-time processing efficiency of the model is as high as 41 frames per second, which makes it ideal for automated inspection of tobacco production lines and effectively solves the problem of tobacco impurity detection.

RESUMEN

This study explored an indoor system for tracking multiple humans and detecting falls, employing three Millimeter-Wave radars from Texas Instruments. Compared to wearables and camera methods, Millimeter-Wave radar is not plagued by mobility inconveniences, lighting conditions, or privacy issues. We conducted an initial evaluation of radar characteristics, covering aspects such as interference between radars and coverage area. Then, we established a real-time framework to integrate signals received from these radars, allowing us to track the position and body status of human targets non-intrusively. Additionally, we introduced innovative strategies, including dynamic Density-Based Spatial Clustering of Applications with Noise (DBSCAN) clustering based on signal SNR levels, a probability matrix for enhanced target tracking, target status prediction for fall detection, and a feedback loop for noise reduction. We conducted an extensive evaluation using over 300 min of data, which equated to approximately 360,000 frames. Our prototype system exhibited a remarkable performance, achieving a precision of 98.9% for tracking a single target and 96.5% and 94.0% for tracking two and three targets in human-tracking scenarios, respectively. Moreover, in the field of human fall detection, the system demonstrates a high accuracy rate of 96.3%, underscoring its effectiveness in distinguishing falls from other statuses.

RESUMEN

Macromolecular crystallography contributes significantly to understanding diseases and, more importantly, how to treat them by providing atomic resolution 3D structures of proteins. This is achieved by collecting X-ray diffraction images of protein crystals from important biological pathways. Spotfinders are used to detect the presence of crystals with usable data, and the spots from such crystals are the primary data used to solve the relevant structures. Having fast and accurate spot finding is essential, but recent advances in synchrotron beamlines used to generate X-ray diffraction images have brought us to the limits of what the best existing spotfinders can do. This bottleneck must be removed so spotfinder software can keep pace with the X-ray beamline hardware improvements and be able to see the weak or diffuse spots required to solve the most challenging problems encountered when working with diffraction images. In this paper, we first present Bragg Spot Detection (BSD), a large benchmark Bragg spot image dataset that contains 304 images with more than 66 000 spots. We then discuss the open source extensible U-Net-based spotfinder Bragg Spot Finder (BSF), with image pre-processing, a U-Net segmentation backbone, and post-processing that includes artifact removal and watershed segmentation. Finally, we perform experiments on the BSD benchmark and obtain results that are (in terms of accuracy) comparable to or better than those obtained with two popular spotfinder software packages (Dozor and DIALS), demonstrating that this is an appropriate framework to support future extensions and improvements.

RESUMEN

This study presents a novel solution for ambient assisted living (AAL) applications that utilizes spiking neural networks (SNNs) and reconfigurable neuromorphic processors. As demographic shifts result in an increased need for eldercare, due to a large elderly population that favors independence, there is a pressing need for efficient solutions. Traditional deep neural networks (DNNs) are typically energy-intensive and computationally demanding. In contrast, this study turns to SNNs, which are more energy-efficient and mimic biological neural processes, offering a viable alternative to DNNs. We propose asynchronous cellular automaton-based neurons (ACANs), which stand out for their hardware-efficient design and ability to reproduce complex neural behaviors. By utilizing the remote supervised method (ReSuMe), this study improves spike train learning efficiency in SNNs. We apply this to movement recognition in an elderly population, using motion capture data. Our results highlight a high classification accuracy of 83.4%, demonstrating the approach's efficacy in precise movement activity classification. This method's significant advantage lies in its potential for real-time, energy-efficient processing in AAL environments. Our findings not only demonstrate SNNs' superiority over conventional DNNs in computational efficiency but also pave the way for practical neuromorphic computing applications in eldercare.

RESUMEN

Real-time heterogeneous parallel embedded digital signal processor (DSP) systems process multiple data streams in parallel in a stringent time interval. This type of system on chip (SoC) requires the network on chip (NoC) to establish multiple symbiotic parallel data transmission paths with ultra-low transmission latency in real time. Our early NoC research PCCNOC meets this need. The PCCNOC uses packet routing to establish and lock a transmission circuit, so that PCCNOC is perfectly suitable for ultra-low latency and high-bandwidth transmission of long data packets. However, a parallel multi-data stream DSP system also needs to transmit roughly the same number of short data packets for job configuration and job execution status reports. While transferring short data packets, the link establishment routing delay of short data packets becomes relatively obvious. Our further research, thus, introduced PaCHNOC, a hybrid NoC in which long data packets are transmitted through a circuit established and locked by routing, and short data packets are attached to the routing packet and the transmission is completed during the routing process, thus avoiding the PCCNOC setup delay. Simulation shows that PaCHNOC performs well in supporting real-time heterogeneous parallel embedded DSP systems and achieves overall latency reduction 65% compared with related works. Finally, we used PaCHNOC in the baseband subsystem of a real 5G base station, which proved that our research is the best NoC for baseband subsystem of 5G base stations, which reduce 31% comprehensive latency in comparison to related works.

RESUMEN

Data acquisition and processing for cryo-electron tomography can be a significant bottleneck for users. To simplify and streamline the cryo-ET workflow, Tomo Live, an on-the-fly solution that automates the alignment and reconstruction of tilt-series data, enabling real-time data-quality assessment, has been developed. Through the integration of Tomo Live into the data-acquisition workflow for cryo-ET, motion correction is performed directly after each of the acquired tilt angles. Immediately after the tilt-series acquisition has completed, an unattended tilt-series alignment and reconstruction into a 3D volume is performed. The results are displayed in real time in a dedicated remote web platform that runs on the microscope hardware. Through this web platform, users can review the acquired data (aligned stack and 3D volume) and several quality metrics that are obtained during the alignment and reconstruction process. These quality metrics can be used for fast feedback for subsequent acquisitions to save time. Parameters such as Alignment Accuracy, Deleted Tilts and Tilt Axis Correction Angle are visualized as graphs and can be used as filters to export only the best tomograms (raw data, reconstruction and intermediate data) for further processing. Here, the Tomo Live algorithms and workflow are described and representative results on several biological samples are presented. The Tomo Live workflow is accessible to both expert and non-expert users, making it a valuable tool for the continued advancement of structural biology, cell biology and histology.

Asunto(s)

RESUMEN

Large medical imaging data sets are becoming increasingly available. A common challenge in these data sets is to ensure that each sample meets minimum quality requirements devoid of significant artefacts. Despite a wide range of existing automatic methods having been developed to identify imperfections and artefacts in medical imaging, they mostly rely on data-hungry methods. In particular, the scarcity of artefact-containing scans available for training has been a major obstacle in the development and implementation of machine learning in clinical research. To tackle this problem, we propose a novel framework having four main components: (1) a set of artefact generators inspired by magnetic resonance physics to corrupt brain MRI scans and augment a training dataset, (2) a set of abstract and engineered features to represent images compactly, (3) a feature selection process that depends on the class of artefact to improve classification performance, and (4) a set of Support Vector Machine (SVM) classifiers trained to identify artefacts. Our novel contributions are threefold: first, we use the novel physics-based artefact generators to generate synthetic brain MRI scans with controlled artefacts as a data augmentation technique. This will avoid the labour-intensive collection and labelling process of scans with rare artefacts. Second, we propose a large pool of abstract and engineered image features developed to identify 9 different artefacts for structural MRI. Finally, we use an artefact-based feature selection block that, for each class of artefacts, finds the set of features that provide the best classification performance. We performed validation experiments on a large data set of scans with artificially-generated artefacts, and in a multiple sclerosis clinical trial where real artefacts were identified by experts, showing that the proposed pipeline outperforms traditional methods. In particular, our data augmentation increases performance by up to 12.5 percentage points on the accuracy, F1, F2, precision and recall. At the same time, the computation cost of our pipeline remains low - less than a second to process a single scan - with the potential for real-time deployment. Our artefact simulators obtained using adversarial learning enable the training of a quality control system for brain MRI that otherwise would have required a much larger number of scans in both supervised and unsupervised settings. We believe that systems for quality control will enable a wide range of high-throughput clinical applications based on the use of automatic image-processing pipelines.

Asunto(s)

RESUMEN

In recent years, the convergence of edge computing and sensor technologies has become a pivotal frontier revolutionizing real-time data processing. In particular, the practice of data acquisition-which encompasses the collection of sensory information in the form of images and videos, followed by their transmission to a remote cloud infrastructure for subsequent analysis-has witnessed a notable surge in adoption. However, to ensure seamless real-time processing irrespective of the data volume being conveyed or the frequency of incoming requests, it is vital to proactively locate resources within the cloud infrastructure specifically tailored to data-processing tasks. Many studies have focused on the proactive prediction of resource demands through the use of deep learning algorithms, generating considerable interest in real-time data processing. Nonetheless, an inherent risk arises when relying solely on predictive resource allocation, as it can heighten the susceptibility to system failure. In this study, a framework that includes algorithms that periodically monitor resource requirements and dynamically adjust resource provisioning to match the actual demand is proposed. Under experimental conditions with the Bitbrains dataset, setting the network throughput to 300 kB/s and with a threshold of 80%, the proposed system provides a 99% performance improvement in terms of the autoscaling algorithm and requires only 0.43 ms of additional computational overhead compared to relying on a simple prediction model alone.

RESUMEN

Korean words like balgda 'bright/become bright' and gilda 'long/become long' are categorially ambiguous; they can appear as both adjectives and verbs. Some suggest that these words are listed under separate lexical entries, while others propose that they share one single lexical entry, and that the verb form is morphologically derived from the base adjective through a process called zero derivation. This study presents the results of a real-time experiment that investigates whether these words involve zero derivation and if so, how zero derivation may affect the real-time processing of these words. Our findings suggest that the reader recognizes the base adjective and obtains the derived-verb form by virtue of adding a covert category-changing morpheme in real-time sentence processing. This study provides promising evidence of the zero derivation of Korean categorially ambiguous adjectives and verbs, as well as crosslinguistic evidence of the role of covert structure in lexical access.

RESUMEN

Automatic hand gesture recognition in video sequences has widespread applications, ranging from home automation to sign language interpretation and clinical operations. The primary challenge lies in achieving real-time recognition while managing temporal dependencies that can impact performance. Existing methods employ 3D convolutional or Transformer-based architectures with hand skeleton estimation, but both have limitations. To address these challenges, a hybrid approach that combines 3D Convolutional Neural Networks (3D-CNNs) and Transformers is proposed. The method involves using a 3D-CNN to compute high-level semantic skeleton embeddings, capturing local spatial and temporal characteristics of hand gestures. A Transformer network with a self-attention mechanism is then employed to efficiently capture long-range temporal dependencies in the skeleton sequence. Evaluation of the Briareo and Multimodal Hand Gesture datasets resulted in accuracy scores of 95.49% and 97.25%, respectively. Notably, this approach achieves real-time performance using a standard CPU, distinguishing it from methods that require specialized GPUs. The hybrid approach's real-time efficiency and high accuracy demonstrate its superiority over existing state-of-the-art methods. In summary, the hybrid 3D-CNN and Transformer approach effectively addresses real-time recognition challenges and efficient handling of temporal dependencies, outperforming existing methods in both accuracy and speed.

Asunto(s)

RESUMEN

The final objective of the study herein reported is the preliminary evaluation of the capability of an original, real-time SHM system applied to a full-scale wing-box section as a significant aircraft component, during an experimental campaign carried out at the Piaggio Lab in Villanova D'Albenga, Italy. In previous works, the authors have shown that such a system could be applied to composite beams, to reveal damage along the bonding line between a longitudinal stiffening element and the cap. Utilizing a suitable scaling process, such work has then been exported to more complex components, in order to confirm the outcomes that were already achieved, and, possibly, expanding the considerations that should drive the project towards an actual implementation of the proposed architecture. Relevant topics dealt with in this publication concern the application of the structural health monitoring system to different temperature ranges, by taking advantage of a climatic room operating at the Piaggio sites, and the contemporary use of several algorithms for real-time elaborations. Besides the real-time characteristics already introduced and discussed previously, such further steps are essential for applying the proposed architecture on board an aircraft, and to increase reliability aspects by accessing the possibility of comparing different information derived from different sources. The activities herein reported have been carried out within the Italian segment of the RESUME project, a joint co-operation between the Ministry of Defense of Israel and the Ministry of Defense of Italy.

RESUMEN

Smart living, a concept that has gained increasing attention in recent years, revolves around integrating advanced technologies in homes and cities to enhance the quality of life for citizens. Sensing and human action recognition are crucial aspects of this concept. Smart living applications span various domains, such as energy consumption, healthcare, transportation, and education, which greatly benefit from effective human action recognition. This field, originating from computer vision, seeks to recognize human actions and activities using not only visual data but also many other sensor modalities. This paper comprehensively reviews the literature on human action recognition in smart living environments, synthesizing the main contributions, challenges, and future research directions. This review selects five key domains, i.e., Sensing Technology, Multimodality, Real-time Processing, Interoperability, and Resource-Constrained Processing, as they encompass the critical aspects required for successfully deploying human action recognition in smart living. These domains highlight the essential role that sensing and human action recognition play in successfully developing and implementing smart living solutions. This paper serves as a valuable resource for researchers and practitioners seeking to further explore and advance the field of human action recognition in smart living.

Asunto(s)

RESUMEN

Deep learning has a better output quality compared with traditional algorithms for video super-resolution (SR), but the network model needs large resources and has poor real-time performance. This paper focuses on solving the speed problem of SR; it achieves real-time SR by the collaborative design of a deep learning video SR algorithm and GPU parallel acceleration. An algorithm combining deep learning networks with a lookup table (LUT) is proposed for the video SR, which ensures both the SR effect and ease of GPU parallel acceleration. The computational efficiency of the GPU network-on-chip algorithm is improved to ensure real-time performance by three major GPU optimization strategies: storage access optimization, conditional branching function optimization, and threading optimization. Finally, the network-on-chip was implemented on a RTX 3090 GPU, and the validity of the algorithm was demonstrated through ablation experiments. In addition, SR performance is compared with existing classical algorithms based on standard datasets. The new algorithm was found to be more efficient than the SR-LUT algorithm. The average PSNR was 0.61 dB higher than the SR-LUT-V algorithm and 0.24 dB higher than the SR-LUT-S algorithm. At the same time, the speed of real video SR was tested. For a real video with a resolution of 540×540, the proposed GPU network-on-chip achieved a speed of 42 FPS. The new method is 9.1 times faster than the original SR-LUT-S fast method, which was directly imported into the GPU for processing.

RESUMEN

Pronoun interpretation is often described as relying on a comprehender's mental model of discourse. For example, in some psycholinguistic accounts, interpreting pronouns involves a process of retrieval, whereby a pronoun is resolved by accessing information from its linguistic antecedent. However, linguistic antecedents are neither necessary nor sufficient for interpreting a pronoun, and even when an antecedent has been introduced in earlier discourse, there is little evidence for the retrieval of linguistic form. The current study extends our understanding of pronoun interpretation by examining whether the semantics of antecedent expressions are retrieved from representations of past discourse. Participants were instructed to move displayed objects in a Visual World eye-tracking task. In some cases, the semantics of the antecedent were no longer viable after an instruction was completed (e.g., "Move the house on the left to area 12," where the result was that a different house is now the leftmost one). In this case, retrieving antecedent semantics at the point of hearing a subsequent pronoun ("Now, move it…") should entail a processing penalty. Instead, the results showed that antecedent semantics have no direct effect on interpretation, raising additional questions about the role that retrieval might play in pronoun interpretation.

Asunto(s)

RESUMEN

Smart metering systems development and implementation in power distribution networks can be seen as an important factor that led to a major technological upgrade and one of the first steps in the transition to smart grids. Besides their main function of power consumption metering, as is demonstrated in this work, the extended implementation of smart metering can be used to support many other important functions in the electricity distribution grid. The present paper proposes a new solution that uses a frequency feature-based method of data time-series provided by the smart metering system to estimate the energy contour at distribution level with the aim of improving the quality of the electricity supply service, of reducing the operational costs and improving the quality of electricity measurement and billing services. The main benefit of this approach is determining future energy demand for optimal energy flow in the utility grid, with the main aims of the best long term energy production and acquisition planning, which lead to lowering energy acquisition costs, optimal capacity planning and real-time adaptation to the unpredicted internal or external electricity distribution branch grid demand changes. Additionally, a contribution to better energy production planning, which is a must for future power networks that benefit from an important renewable energy contribution, is intended. The proposed methodology is validated through a case study based on data supplied by a real power grid from a medium sized populated European region that has both economic usage of electricity-industrial or commercial-and household consumption. The analysis performed in the proposed case study reveals the possibility of accurate energy contour forecasting with an acceptable maximum error. Commonly, an error of 1% was obtained and in the case of the exceptional events considered, a maximum 15% error resulted.

RESUMEN

The present paper reports the outcomes of activities concerning a real-time SHM system for debonding flaw detection based on ground testing of an aircraft structural component as a basis for condition-based maintenance. In this application, a damage detection method unrelated to structural or load models is investigated. In the reported application, the system is applied for real-time detection of two flaws, kissing bond type, artificially deployed over a full-scale composite spar under the action of external bending loads. The proposed algorithm, local high-edge onset (LHEO), detects damage as an edge onset in both the space and time domains, correlating current strain levels to next strain levels within a sliding inner product proportional to the sensor step and the acquisition time interval, respectively. Real-time implementation can run on a consumer-grade computer. The SHM algorithm was written in Matlab and compiled as a Python module, then called from a multiprocess wrapper code with separate operations for data reception and data elaboration. The proposed SHM system is made of FBG arrays, an interrogator, an in-house SHM code, an original decoding software (SW) for real-time implementation of multiple SHM algorithms and a continuous interface with an external operator.

Asunto(s)

RESUMEN

Synthetic aperture radar (SAR), which can generate images of regions or objects, is an important research area of radar. The chirp scaling algorithm (CSA) is a representative SAR imaging algorithm. The CSA has a simple structure comprising phase compensation and fast Fourier transform (FFT) operations by replacing interpolation for range cell migration correction (RCMC) with phase compensation. However, real-time processing still requires many computations and a long execution time. Therefore, it is necessary to develop a hardware accelerator to improve the speed of algorithm processing. In addition, the demand for a small SAR system that can be mounted on a small aircraft or drone and that satisfies the constraints of area and power consumption is increasing. In this study, we proposed a CSA-based SAR processor that supports FFT and phase compensation operations and presents field-programmable gate array (FPGA)-based implementation results. We also proposed a modified CSA flow that simplifies the traditional CSA flow by changing the order in which the transpose operation occurs. Therefore, the proposed CSA-based SAR processor was designed to be suitable for modified CSA flow. We designed the multiplier for FFT to be shared for phase compensation, thereby achieving area efficiency and simplifying the data flow. The proposed CSA-based SAR processor was implemented on a Xilinx UltraScale+ MPSoC FPGA device and designed using Verilog-HDL. After comparing the execution times of the proposed SAR processor and the ARM cortex-A53 microprocessor, we observed a 136.2-fold increase in speed for the 4096 × 4096-pixel image.

Asunto(s)