A lightweight and real-time surgical action detection framework using multi-contextual and decoupled representations

Generating Rare Surgical Events Using CycleGAN: Addressing Lack of Data for Artificial Intelligence Event Recognition

In the actual industrial production environment, the surface defects of products are subtle, and the number of different types of defect data samples is also quite small. Most deep learning models rely on a large number of training samples and parameters to achieve high-precision defect detection. At the same time, the edge computing layer in the actual industrial environment may also encounter transmission delays and insufficient resources. Training a proper model for a specific type of surface defect while simultaneously satisfying the real-time accuracy of defect detection is still a challenging task. To effectively deal with the above challenges, we propose an edge-cloud computing defect detection model based on the intrinsic mean feature detector in the Lie Group space. The modules in the model adopt a symmetrical structure, which can extract related features more effectively. Different from existing models, this model utilizes the Lie Group space intrinsic mean feature as a metric to characterize the essential attributes of different types of surface defects. In addition, we propose an intrinsic mean attention mechanism in the Lie Group manifold space that is easy to implement at the edge service layer without increasing the number of model parameters, thereby enhancing the detection performance of tiny surface defects. Extensive experiments on three publicly available and challenging datasets reveal the superiority of our model in terms of detection accuracy, real-time detection, number of parameters, and computational performance. In addition, our proposed model also shows competitiveness and advantages compared with state-of-the-art models.

How to get the potential interaction intent is the key to design a personalized intelligent kitchen assistance system based on Human Robot Interaction (HRI), which can facilitate the elder people in their daily life. Eye movement includes abundant user intent information, eye tracking technology can be used to obtain and analyze this information, thus identifying user’s potential intent. This paper proposed an eye tracking based object recognition method, which could recognize objects that the user was watching. Firstly, multiple fixation points were clustered into a fixation point that best represented the user’s intent by research on the study of the clustering algorithm. Then the object recognition algorithm based on PP-ShiTu platform was used to obtain the specific location of the object in the scene image. Finally, this paper proposed a fixation based object recognition module, and used the module to design and implement a personalized intelligent kitchen assistance system, and verified the effectiveness through experiments. The experimental results showed that when there was only single food recognition, the average recognition accuracy of the personalized intelligent kitchen assistance system reached over 80%, and the average time delay of the system was about 1400 ms.

The intelligent judicial trial assistance system is an intelligent auxiliary judicial trial tool based on generative artificial intelligence and machine learning technology. It can achieve auxiliary functions such as case upload, trial result prediction, and legal document generation through automated, digital, and informational means. It utilizes technologies such as Recurrent Neural Networks (RNN) and Long Short-Term Memory networks (LSTM) to train generative models for generating trial recommendations and assisting in decision-making. Through in-depth research on the working principles and optimization paths of the intelligent judicial trial assistance system, the best applicable path for the system in judicial trial practice is explored. By establishing optimized models, risks and biases are minimized to the greatest extent.

Intelligent assistance systems (IAS) are designed to counteract rising cognitive demands caused by increasingly individualized manufacturing processes in assembly. How IAS affect work characteristics which are crucial for promoting work motivation of employees is yet unclear. Based on the cyber-physical systems transformation framework, the model of routine-biased technological change, and a comprehensive model of work design, we expected in- and decreases in motivational work characteristics (MWC) when working with IAS. Furthermore, we posited a buffering effect of the option of voluntary use on decreasing knowledge characteristics. Applying an online case study with experimental vignette methodology (EVM) allowed us to identify effects of the IAS on MWC before it is widely implemented. 203 German and British blue-collar workers evaluated an assembly workplace according to three experimental conditions (work without IAS, work with IAS, work with voluntary use of IAS). We identified enhanced feedback from the job and information processing in work with IAS in contrast to a traditional assembly workplace but found no restrictions (or elevations) in terms of other task (i.e., autonomy) or knowledge characteristics (i.e., job complexity, problem solving, specialization, skill variety). Thus, our results indicate that the IAS improves some motivational work characteristics of the assembly workplace, although it misses the primary goal of cognitive relief. Our study highlights the need for work design theories that specify the effect of IAS on motivational work characteristics and the potential benefit of IAS in assembly of the future.Supplementary InformationThe online version contains supplementary material available at 10.1007/s10845-023-02086-4.

Temporal graph convolutional network for multi-agent reinforcement learning of action detection

Driver behavior remains a leading factor in road accidents, yet existing monitoring systems typically rely on single data modalities such as facial expressions or speech alone, limiting their reliability and contextual awareness. This work proposes a comprehensive driver behavior monitoring system using multimodal AI, uniquely integrating video, audio, and vehicle speed telemetry — an approach that remains underexplored in existing literature — to predict driver emotions and behaviors in real time. The system analyzes facial cues to detect visual anomalies, processes audio inputs to infer emotional states, and incorporates speed telemetry to provide additional behavioral context. This fusion of modalities is designed to improve classification accuracy and reduce false positives compared to unimodal approaches. Performance evaluation is conducted using benchmark datasets for both video-based and audio-based emotion recognition, with comparative analysis between individual and combined modalities. By addressing the challenges of multimodal integration and real-time processing, this research contributes a novel and effective framework for intelligent driver assistance systems, advancing the goal of enhanced road safety through predictive behavioral intervention. Additionally, this research is being extended to incorporate an intermediate-fusion-based multimodal decision framework, wherein top predictions from image, audio, and vehicle telemetry are jointly fed into a decision model (e.g., Random Forest or SVM) for improved context-aware warning generation. This approach addresses prior limitations of ensemble-style fusion and better aligns with the goals of true multimodal AI.

Navigators for Motion Detection in the 1.5T MRI-linac

Nowadays, early defect detection plays a significant role for the railway safety warning. However, the existing methods cannot satisfy the requirements of real-time and high-precision detection. Here, a high-precision, distributed and on-line method for detecting rail defect is proposed and demonstrated. When a train goes through defects, the instantaneous elastic waves will be excited by the wheel-rail interaction, which will further propagate along railway tracks bidirectionally. Through mounting the backscattering enhanced optical fiber on the railway as sensors, the fiber optic distributed acoustic sensing system can record the propagation trace precisely. Further, the acoustic propagation fitting method is applied onto the propagation data to detect and locate defects along the long-distance railway. Especially, the dual-frequency joint-processing algorithm is proposed to improve the location accuracy. The field test proves that multiple defects along the railway can be successfully identified and located with a standard deviation of 0.314m. To the best of our knowledge, this work is the first report of distributed rail defect detection, which will bring a breakthrough for high-precision structural damage detection in the infrastructures such as the railway, pipeline and tunnel.

Vision is a fundamental sense that profoundly impacts daily life and independence. For visually impaired people (VIP), the absence or impairment of this sense presents significant challenges, particularly in navigating their environment and identifying objects independently. A considerable challenge for visually impaired individuals is the inability to navigate independently and identify objects, which restricts their daily activities and routines. Various investigations have been conducted in the domain of real-time object detection (OD) using deep learning (DL). DL-based techniques are shown to attain performance in OD. In this manuscript, an Intelligent Object Detection-Based Assistive System Using Ivy Optimisation Algorithm (IODAS-IOA) model is proposed for VIPs. The aim is to develop an effective OD system to help visually impaired persons navigate their environment safely and independently. To achieve this, the image pre-processing stage initially employs the Gaussian filtering (GF) method to eliminate noise. Moreover, a novel YOLOv12 method is employed for the OD process. Furthermore, the DenseNet161 method is used for the feature extraction process. Additionally, the bidirectional gated recurrent unit with attention mechanism (BiGRU-AM) method is implemented for classifying the extracted images. Finally, the parameter tuning process is performed using the Ivy optimization algorithm (IOA) method to improve classification performance. Extensive experimentation was performed to validate the performance of the IODAS-IOA approach under the indoor OD dataset. The experimental results of the IODAS-IOA approach emphasized the superior accuracy value of 99.74% over recent techniques.

We aimed to provide causal evidence on the contradictory effects of projection‐based intelligent assistance systems (IASs) for nine motivational work characteristics (MWCs). IASs are increasingly implemented in assembly to counteract rising cognitive workload due to individualized manufacturing processes. However, how IASs enhance or restrict MWCs is largely unknown. We conducted two studies with experimental vignette methodology. In Study 1 (N1 = 169 German employees), we manipulated an assembly workplace (with IAS vs. without IAS) and tested whether findings indicating only positive effects of IASs in the support of a simple assembly process can be transferred to more complex assembly processes. In Study 2 (N2 = 176 German employees), we manipulated again the assembly workplace (with IAS vs. without IAS) and in addition the dynamic of product changes (task rotation after 1 h vs. no task rotation). Analyzing the data with SPSS 27, we found increased feedback from job and information processing and decreased work scheduling, decision‐making, and work methods autonomy when working with IAS. In Study 2, we did not find the main or interaction effects of task rotation on MWCs. Our experimental evidence suggests that working with IASs represents a double‐edged sword regarding MWCs and that the effect of task rotation is limited. Hence, our results provide vital theoretical implications for a much‐needed work design theory that delineates how new technologies shape work design and practical implications for modern assembly.

This research proposes a novel hardware-optimized design to accelerate Convolutional Neural Networks (CNNs) using Verilog HDL. The design is specifically developed for the DARKNET-19 system model, which serves as the backbone of the YOLOv3-tiny algorithm, a widely used framework for real-time object detection in dynamic environments. The CNN architecture was implemented in Verilog HDL and synthesized using Synopsys Design Compiler, with a focus on improving both object detection accuracy and hardware resource efficiency. The proposed design efficiently performs key CNN operations, including convolution, pooling, and activation, enabling faster real-time object detection compared to many existing methods. To improve performance, the hardware design incorporates parallel processing techniques, allowing multiple computations to be executed simultaneously. This significantly reduces the system latency and power consumption. The convolutional layers of the DARKNET-19 architecture are efficiently mapped onto the hardware platform, ensuring optimized data storage and fast memory access, which further enhances processing speed and detection accuracy. An innovative feature of the design is a 2-dimensional image preprocessing module that prepares input images before they are fed into the CNN. This preprocessing stage includes image resizing, brightness normalization, and color adjustment, which helps the CNN process visual data more effectively. After preprocessing, the images pass through several CNN layers. The convolutional layers extract key features from the images, while the pooling and activation layers refine these features to improve detection performance. Finally, the processed data is analyzed by the YOLOv3-tiny algorithm, which identifies and locates objects in the images with high precision. Experimental results demonstrate that the proposed high-speed and resource-efficient hardware architecture is well-suited for real-time object detection applications, particularly in highly dynamic and unpredictable environments.

Legal dispute analysis is crucial for intelligent legal assistance systems. However, current Large Language Models (LLMs) face challenges in understanding complex legal concepts, maintaining reasoning consistency, and accurately citing legal sources. This study presents a framework combining prompt engineering with multi-dimensional knowledge graphs to improve LLM capabilities for legal dispute analysis. The framework comprises a three-stage hierarchical prompt structure (task definition, knowledge background, reasoning guidance) and a three-layer knowledge graph (legal classification ontology layer, representation layer, instance layer). Additionally, four supporting methods enable legal concept retrieval: direct code matching, semantic vector similarity, ontology path reasoning, and professional terminology matching. Systematic testing on 500 test samples integrated from six internationally recognized legal AI benchmark datasets demonstrates performance improvements for mainstream models: F1 score increased from 0.356 to 0.714, BLEU-4 reached 0.451, ROUGE-L F1 improved from 0.34 to 0.71, and legal professional content quality scores increased by 18-20 points (on a 100-point scale). This framework provides a technical approach for legal analysis, contributing to the advancement of intelligent legal assistance systems.

This article considers some problems and features of the application of modern systems of driver assistance on single-track vehicles (“STV”). It is also mentioned the classification of intelligent driver assistance systems and design features of the “STV” that require an improvement for their successful application on such vehicle. There is also described perspective of systems’ development that helps the “STV” driver. Also perspective the “STV” types with high safety construction are considered in this article.

The connected road infrastructure and roadside assistance services constitute an important consumer market segment in the Intelligent Transportation System (ITS) and Smart Cities. A closer look at available such services reveal the presence of data silos, heterogeneity and lack of interoperability. They affect the overall consumer experience and increase the cost of service development & maintenance. This paper proposes an IoT framework for next generation, intelligent roadside assistance system. A data centric architecture is presented along with solutions of the mentioned challenges.