Basic Modules Research Articles

Class semantic enhancement network for semantic segmentation

Existing semantic segmentation methods favor class semantic consistency by extracting long-range contextual features through multi-scale and attention strategies. These methods ignore the relations between feature channels and classes, which are essential to represent consistent class semantics. To this end, we propose the Class Semantic Enhancement Network (CSENet) which boosts the segmentation performance of a backbone network in a coarse-to-fine manner. CSENet consists of two basic modules – (1) the Class Semantic Channel Graph Module (CSCG) module, which captures inter-dependencies among channels and strengthens the channel-class relation, and (2) the Class Prior Fully Convolution (CP-FC) module, which utilizes the channel-class relation as class priors to refine the segmentation results. Extensive experiments have demonstrated that the proposed CSENet is able to learn discriminative feature representations and achieves state-of-the-art performance on three benchmark datasets, including PASCAL Context, ADE20K and COCO Stuff.

A transformer-based dual-domain network for reconstructing FOV extended cone-beam CT images from truncated sinograms in radiation therapy

Background and objectiveCone-beam computed tomography (CBCT) is widely used in clinical radiotherapy, but its small field of view (sFOV) limits its application potential. In this study, a transformer-based dual-domain network (dual_swin), which combined image domain restoration and sinogram domain restoration, was proposed for the reconstruction of complete CBCT images with extended FOV from truncated sinograms. MethodsThe planning CT images with large FOV (LFOV) of 330 patients who received radiation therapy were collected. The synthetic CBCT (sCBCT) images with LFOV were generated from CT images by the trained cycleGAN network, and CBCT images with sFOV were obtained through forward projection, projection truncation, and filtered back projection (FBP), comprising the training and test data. The proposed dual_swin includes sinogram domain restoration, image domain restoration, and FBP layer, and the swin transformer blocks were used as the basic feature extraction module in the network to improve the global feature extraction ability. The proposed dual_swin was compared with the image domain method, the sinogram domain method, the U-Net based dual domain network (dual_Unet), and the traditional iterative reconstruction method based on prior image and conjugate gradient least-squares (CGLS) in the test of sCBCT images and clinical CBCT images. The HU accuracy and body contour accuracy of the predicted images by each method were evaluated. ResultsThe images generated using the CGLS method were fuzzy and obtained the lowest structural similarity (SSIM) among all methods in the test of sCBCT and clinical CBCT images. The predicted images by the image domain methods are quite different from the ground truth and have low accuracy on HU value and body contour. In comparison with image domain methods, sinogram domain methods improved the accuracy of HU value and body contour but introduced secondary artifacts and distorted bone tissue. The proposed dual_swin achieved the highest HU and contour accuracy with mean absolute error (MAE) of 23.0 HU, SSIM of 95.7%, dice similarity coefficient (DSC) of 99.6%, and Hausdorff distance (HD) of 4.1 mm in the test of sCBCT images. In the test of clinical patients, images that were predicted by dual_swin yielded MAE, SSIM, DSC, and HD of 38.2 HU, 91.7%, 99.0%, and 5.4 mm, respectively. The predicted images by the proposed dual_swin has significantly higher accuracy than the other methods (P < 0.05). ConclusionsThe proposed dual_swin can accurately reconstruct FOV extended CBCT images from the truncated sinogram to improve the application potential of CBCT images in radiotherapy.

Reconstructed SqueezeNext with C-CBAM for offline handwritten Chinese character recognition.

Handwritten Chinese character recognition (HCCR) is a difficult problem in character recognition. Chinese characters are diverse and many of them are very similar. The HCCR model consumes a large number of computational resources during runtime, making it difficult to deploy to resource-limited development platforms. In order to reduce the computational consumption and improve the operational efficiency of such models, an improved lightweight HCCR model is proposed in this article. We reconstructed the basic modules of the SqueezeNext network so that the model would be compatible with the introduced attention module and model compression techniques. The proposed Cross-stage Convolutional Block Attention Module (C-CBAM) redeploys the Spatial Attention Module (SAM) and the Channel Attention Module (CAM) according to the feature map characteristics of the deep and shallow layers of the model, targeting enhanced information interaction between the deep and shallow layers. The reformulated intra-stage convolutional kernel importance assessment criterion integrates the normalization nature of the weights and allows for structured pruning in equal proportions for each stage of the model. The quantization aware training is able to map the 32-bit floating-point weights in the pruned model to 8-bit fixed-point weights with minor loss. Pruning with the new convolutional kernel importance evaluation criterion proposed in this article can achieve a pruning rate of 50.79% with little impact on the accuracy rate. The various optimization methods can compress the model to 1.06 MB and achieve an accuracy of 97.36% on the CASIA-HWDB dataset. Compared with the initial model, the volume is reduced by 87.15%, and the accuracy is improved by 1.71%. The model proposed in this article greatly reduces the running time and storage requirements of the model while maintaining accuracy.

LCNNet: Light-weight convolutional neural networks for understanding the highly congested scenes

With the development of convolutional neural networks, many improved algorithms have been successively proposed to promote the accuracy of dense crowd counting. However, these algorithms are deployed with expensive computing resources, which is unbearable for small devices such as embedded systems with limited computing resources. To realize the real-time counting on the small devices, it is of great significance how to trade off the computation cost and processing accuracy of the dense crowd-counting algorithm. Thus, we propose a lightweight dense crowd-counting algorithm (LCNNet) to improve this issue. Specifically, the proposed LCNNet consists of two subnetworks, a feature extraction subnetwork, and a regression subnetwork, with a bottleneck depth-separable convolution with a residuals module as the basic module. The LCNNet effectively improves computational efficiency and reduces the computational cost, which can be performed on small devices. Extensive evaluations on four benchmark datasets well demonstrate the effectiveness of the proposed LCNNet for dense crowd-counting models. Meanwhile, the proposed LCNNet can maintain a comparable level of computational accuracy and computational cost on Vehicle counting datasets.

The Mobile Augmented Reality Application for Improving Learning of Electronic Component Module in TVET

Teens and young adults may get training in anything from the basics to advanced skills in various workplace and academic settings at Technical and Vocational Education Training and Education (TVET) institutions. Some aspects of teaching and learning in TVET cannot be articulated clearly, and trainees cannot perceive how things fit together. The study was conducted to determine the optimal platform to develop mobile Augmented Reality applications for TVET trainees and, to assess the TVET trainee’s readiness for AR-based mobile application training deployment. An online questionnaire was sent to trainees at Industrial Training Institute in Malaysia via the online system. A marker-based Augmented Reality application was created for the Basic Electronic Components module utilizing Unity software, the Vuforia engine, and C# script. Finally, the trainees were allowed to test the generated application. The trainees were interviewed to obtain data on their responses. The results indicate that 83% of the TVET trainees own and use android as the application platform. The results of the pre-test and post-tests used to gauge the success of the Augmented Reality application show that its usage in the sub-learning module significantly improved memory recalls for the TVET trainees. The outcomes showed that the Augmented Reality application suited the participants' learning needs and improved the effectiveness of their learning. The result from this project will serve as a pre-test for determining the most suitable platform to deploy the Augmented Reality application to be developed in the future.

VLDNet: An Ultra-Lightweight Crop Disease Identification Network

Existing deep learning methods usually adopt deeper and wider network structures to achieve better performance. However, we found that this rule does not apply well to crop disease identification tasks, which inspired us to rethink the design paradigm of disease identification models. Crop diseases belong to fine-grained features and lack obvious patterns. Deeper and wider network structures will cause information loss of features, which will damage identification efficiency. Based on this, this paper designs a very lightweight disease identification network called VLDNet. The basic module VLDBlock of VLDNet extracts intrinsic features through 1 × 1 convolution, and uses cheap linear operations to supplement redundant features to improve feature extraction efficiency. In inference, reparameterization technology is used to further reduce the model size and improve inference speed. VLDNet achieves state-of-the-art model (SOTA) latency-accuracy trade-offs on self-built and public datasets, such as equivalent performance to Swin-Tiny with a parameter size of 0.097 MB and 0.04 G floating point operations (FLOPs), while reducing parameter size and FLOPs by 297 times and 111 times, respectively. In actual testing, VLDNet can recognize 221 images per second, which is far superior to similar accuracy models. This work is expected to further promote the application of deep learning-based crop disease identification methods in practical production.

Module Development on Basic Laws of Chemistry Based on the 5E Instructional Model to Improve Science Process Skills in Senior High School

The module is one of the teaching materials that is arranged as a whole and systematically. Modules based on learning models can change learning patterns from teacher center to student center. One of the learning models developed specifically for subject matter in the field of science is 5E instructional. The 5E Instructional Model consists of 5 stages namely engagement, exploration, explanation, elaboration and evaluation. The purpose of this study was to develop an instructional module based on 5E chemical laws to improve the process skills of science students. The type of research conducted is research &amp; development (R&amp;D) with the ADDIE development model. The results obtained for the validity of the construct and content were 0.89 and 0.91 with the valid category. Based on the results of the teacher and student response questionnaires, an average practicality score of 86.96% and 86.10% was obtained in the very practical category. The results of the effectiveness test conducted in the experimental and control classes showed that the N-gain value for the experimental class was higher than the control class, namely 0.84 and 0.79. Based on the results of the t-test on student learning outcomes, it was found that tt &lt; th means that the learning outcomes of classes using the 5E instructional module were significantly higher compared to classes that did not use the 5E instructional module. Therefore, it can be said that the 5E-based basic chemistry law module is valid, practical, and effective. The effectiveness of the developed module is seen from the learning outcomes of students and science process skills, so that this module can already be used in the learning process.

Distributed Ledger as a Service: A Web 3.0-Oriented Architecture

This paper proposes a general and interoperable Web of Things (WoT)-oriented architecture to support a distributed storage application. In particular, the focus is on a distributed ledger service dedicated to machine-to-machine (M2M) transactions occurring in an intelligent ecosystem. For this purpose, the basic functional modules have been characterized and integrated into a comprehensive framework relying on an IOTA approach. Furthermore, a general protocol that is built upon an underlying publish-and-subscribe framework is proposed to support all the application phases. The proposed approach has been validated by a simulation campaign targeting the achievable latency and throughput and, further, by a qualitative analysis of high-level metrics, both pointing out several advantages in terms of interoperability, scalability, and mobility support, together with addressing some constraints affecting service availability and security.

Design, implementation, and analysis of an asymmetrically configured DK-type multilevel inverter

ABSTRACT This work proposes a new configuration of asymmetrical MLI which is called a dual kite type (DK-type) multilevel inverter. Its special features include the low total standing voltage (TSV) along with less number of total components. The basic module of the proposed structure is able to generate 15 levels without the use of h-bridge. This results in decreasing the voltage stress considerably. The extension of the proposed circuit can be done in order to increase the levels. The operation of proposed MLI has been discussed in detail, and its comparison with some recently proposed topologies has been done in the paper. Moreover, the simulation results for different constant and dynamic loading conditions have been presented and experimental verification of the same has been done and results have been shown. Fundamental frequency-based switching strategy, also called nearest level control (NLC) modulation technique, is used for providing the switching signals. Further, loss evaluation in the proposed structure has been completed with the help of PLECS software. 5.5% THD is present in the output voltage. The proposed MLI has the maximum efficiency of 98.68% at the output power level of 0.5 kW. The results and analysis presented in the paper describe the improved performance of the proposed MLI.

Power generation and wave attenuation of a hybrid system involving a heaving cylindrical wave energy converter in front of a parabolic breakwater

Deploying a cylindrical heaving wave energy converter (WEC) in front of a parabolic breakwater forms a basic module for synergetic coast protection and power generation. The interactions in the hybrid system are important to its performance but poorly understood. Here the power amplification effect of the parabolic breakwater on the WEC and the additional wave attenuation effect of the WEC on the parabolic breakwater are investigated based on the potential flow theory of linear regular waves. A parabolic breakwater and five WECs with optimized geometry and generator parameters are employed in comparative numerical studies conducted using an open-source code HAMS. Results show that the parabolic breakwater has similar power amplification effects on different WECs, despite the significant differences in their dimensions and power-take-off damping. The power of a WEC can be increased by more than 120% at its natural period. A WEC can shadow the region behind it and reduce the wave amplitude on the opening wall of the parabolic breakwater. It also slightly reduces the amplitude of the relatively high waves in the protection zone behind the breakwater. A flatter WEC with a larger diameter-to-draft ratio is recommended as it generates more power and has a stronger shadow effect.

Programmable In-Memory Computing Circuit for Solving Combinatorial Matrix Operation in One Step

Matrix operations are widely used in practical engineering, but the traditional processing methods rely on the loop iterations and neural network algorithm on the software, requiring a long time to calculate. To address such problem, this paper proposes full hardware in-memory computing circuits based on programmable memristor unit array that can solve combinatorial matrix operations of any order in just one step. First, two basic circuit modules are introduced, which can respectively solve matrix multiplication and matrix equation. Further, the basic modules can be linked to solve combinatorial matrix operations with different forms. It’s worth noting that every module can parallel program the value of each memristor in the memristor unit array and complete one-step computation by hardware. Then, some matrix operations are given in the paper as examples to prove the high accuracy of proposed method, where the average accuracy rate achieves <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$99\%$</tex-math> </inline-formula> . The PSPICE simulation results demonstrate that the processing speed is improved enormously according to the comparison of hardware and software. Moreover, the proposed method has broad application prospect in practical engineering, such as using designed combinational circuit to solve domain shift problem in zero-shot learning, which greatly accelerates the training process of zero-shot learning.

A novel reduced instruction set computer-communication processor design using field programmable gate array

In this paper, a novel reduced instruction set computer (RISC)- communication processor (RCP) has been designed with 32-bit operations which access 64-bit instruction format and implemented using field programmable gate array (FPGA). The design of the RISC processor is facilitated with communication operations like basic signals sine, cosine, and square, and modulation schemes like amplitude modulation, amplitude shift keying, analog, and digital quadrature amplitude modulation. Additionally, application-oriented operations like a traffic light, digital clock, and linear feedback shift register are included in the design. The pipeline mechanism is incorporated in the design to enhance the performance characteristics of the processor, hence allowing the execution of the instructions more effectively. Also, the design is implemented with Xilinx Virtex 7 family FPGA. The device utilization analysis of the proposed FPGA along with different FPGA families is evaluated and compared.

Backstepping Synchronization Control for Three-dimensional Chaotic Oscillatory System via DNA Strand Displacement.

In previous work, a large number of articles have been published to implement coupled synchronization of chaotic systems in DNA-based reaction networks. Up to now, there have been few studies on backstepping synchronous control of chaotic systems through DNA strand displacement. A backstepping synchronization control approach for three-dimensional chaotic system by employing DNA strand displacement is developed in this study. First of all, using the programming properties of DNA molecules, four basic strand displacement reaction modules are given. In the light of these reaction modules as well as the law of mass action kinetics, a novel three-dimensional DNA chaotic system is presented. Second of all, by relying on backstepping control theory and DNA reaction modules, three synchronous controllers are developed to guarantee the synchronization between two three-dimensional DNA chaotic systems. Final of all, numerical simulation results are carried out to verify the validity and applicability of the backstepping synchronization control.

Bone/cartilage organoid on-chip: Construction strategy and application.

The necessity of disease models for bone/cartilage related disorders is well-recognized, but the barrier between ex-vivo cell culture, animal models and the real human body has been pending for decades. The organoid-on-a-chip technique showed opportunity to revolutionize basic research and drug screening for diseases like osteoporosis and arthritis. The bone/cartilage organoid on-chip (BCoC) system is a novel platform of multi-tissue which faithfully emulate the essential elements, biologic functions and pathophysiological response under real circumstances. In this review, we propose the concept of BCoC platform, summarize the basic modules and current efforts to orchestrate them on a single microfluidic system. Current disease models, unsolved problems and future challenging are also discussed, the aim should be a deeper understanding of diseases, and ultimate realization of generic ex-vivo tools for further therapeutic strategies of pathological conditions.

Symmetrical Cross-Connected T-Type Multilevel Inverter With Reduce Device Count

This paper presents a symmetrical, cross-connected T-type (CT-type) multilevel inverter (MLI) configuration with reduced device count (RDC). It comprises eight power switches, two DC sources, and four capacitors with self-voltage balancing. The generalized topology is formed by two CT-type basic modules connected by two power switches. These topologies operate using DC sources and capacitors having the same or different voltage levels, which can be extended by cascading the modules to attain a higher voltage level suitable for high-power applications. Therefore, the proposed CT-Type MLI is a good choice for renewable energy applications that require a lower input voltage source magnitude to attain high voltage levels. This paper uses the Matlab/Simulink platform to simulate the proposed CT-type topology using the level-shifted pulse width modulation (LS-PWM) technique. Further, a laboratory prototype is developed to test and validate the feasibility and effectiveness of the proposed MLI. Finally, a comparative analysis of the number of DC sources, capacitors, power switches, drivers, and total standing voltage (TSV) is presented against similar recent topologies.

Using a q-shuffle algebra to describe the basic module V(Λ_0) for the quantized enveloping algebra Uq(sl^2)

We consider the quantized enveloping algebra Uq(sl^2). and its basic module V(Λ0). This module is infinite-dimensional, irreducible, integrable, and highest-weight. We describe V(Λ0) using a q-shuffle algebra in the following way. Start with the free associative algebra V on two generators x, y. The standard basis for V consists of the words in x, y. In 1995 M. Rosso introduced an associative algebra structure on V, called a q-shuffle algebra. For u, v ∈ {x, y} their q-shuffle product is u ⋆ v = uv + q(u,v)vu, where (u,v) = 2 (resp. (u,v) = − 2) if u = v (resp. u ≠ v). Let U denote the subalgebra of the q-shuffle algebra V that is generated by x, y. Rosso showed that the algebra U is isomorphic to the positive part of Uq(sl^2). In our first main result, we turn U into a Uq(sl^2).-module. Let U denote the Uq(sl^2).-submodule of U generated by the empty word. In our second main result, we show that the Uq(sl^2).-modules U and V(Λ0) are isomorphic. Let V denote the subspace of V spanned by the words that do not begin with y or xx. In our third main result, we show that U = U ∩ V.

Dynamic DNA nanomachines for amplification imaging of diseased cells based on stimuli-responsive mechanism

As a basic functional unit, living cell with sophisticated structures play an indispensable role in life activities. Since the abnormality of important molecules inside cells is closely related to diseases, the dynamic analysis and spatio-temporal monitoring of specific molecules in living cells can provide precious information for the diagnosis and treatment of diseases. More recently, DNA has not only been recognized as the carrier of genetic information, but has also used as a robust building block for the assembly of multitudinous nanoscale structures due to the intrinsic advantages of high programmability of classic Watson–Crick base-pairing rule. Intensive study promotes the rapid progress of nanotechnology in various fields, such as bioimaging, diagnosis, and therapeutics. Among numerous well-defined DNA nanomaterials, DNA nanomachines have been widely exploited in cell imaging owing to their desirable ability to achieve high-resolution temporal and spatial images in response to endogenous or exogenous stimuli. In brief, elaborate DNA nanomachines can undergo structural changes upon the stimuli of target analytes or environmental factors, resulting in rapid increase or reduction of output signals and thereby indirectly reflecting the expression level of targets. DNA nanomachines with high sensitivity and specificity contribute to the recognition of diseased tissues. In this review, we introduce the basic assembly modules of DNA nanomachines and summarize the recent advances in dynamic DNA nanomachines for diseased-cell imaging. Finally, the current challenges and future directions of DNA nanomachines for bioimaging are discussed.

A Soft Reconfigurable Circulator Enabled by Magnetic Liquid Metal Droplet for Multifunctional Control of Soft Robots.

Integrated control circuits with multiple computation functions are essential for soft robots to achieve diverse complex real tasks. However, designing compliant yet simple circuits to embed multiple computation functions in soft electronic systems above the centimeter scale is still a tough challenge. Herein, utilizing smooth cyclic motions of magnetic liquid metal droplets (MLMD) in specially designed and surface-modified circulating channels, a soft reconfigurable circulator (SRC) consisting of three simple and reconfigurable basic modules is described. Through these modules, MLMD can utilize their conductivity and extreme deformation capabilities to transfer their simple cyclic motions as input signals to programmable electrical output signals carrying computing information. The obtained SRCs make it possible for soft robots to perform complex computing tasks, such as logic, programming, and self-adaptive control (a combination of programming and feedback control). Following, a digital logic-based grasping function diagnosis, a locomotion reprogrammable soft car, and a self-adaptive control-based soft sorting gripper are demonstrated to verify SRCs' capabilities. The unique attributes of MLMD allow complex computations based on simple configurations and inputs, which provide new ways to enhance soft robots' computing capabilities.

SheepFaceNet: A Speed-Accuracy Balanced Model for Sheep Face Recognition.

The recognition of sheep faces based on computer vision has improved the efficiency and effectiveness of individual sheep identification, providing technical support for the development of smart farming. However, current recognition models have problems such as large parameter sizes, slow recognition speed, and difficult deployment. Therefore, this paper proposes an efficient and fast basic module called Eblock and uses it to build a lightweight sheep face recognition model called SheepFaceNet, which achieves the best balance between speed and accuracy. SheepFaceNet includes two modules: SheepFaceNetDet for detection and SheepFaceNetRec for recognition. SheepFaceNetDet uses Eblock to construct the backbone network to enhance feature extraction capability and efficiency, designs a bidirectional FPN layer (BiFPN) to enhance geometric location ability, and optimizes the network structure, which affects inference speed, to achieve fast and accurate sheep face detection. SheepFaceNetRec uses Eblock to construct the feature extraction network, uses ECA channel attention to improve the effectiveness of feature extraction, and uses multi-scale feature fusion to achieve fast and accurate sheep face recognition. On our self-built sheep face dataset, SheepFaceNet recognized 387 sheep face images per second with an accuracy rate of 97.75%, achieving an advanced balance between speed and accuracy. This research is expected to further promote the application of deep-learning-based sheep face recognition methods in production.

Research on the Lane Recognition Method Based on Computer Vision

The lane line is the most important traffic sign in road traffic and plays a significant function in restricting and guaranteeing the running of vehicles. Whether in the vehicle safety driving system or in the intelligent vehicle navigation based on machine vision, lane detection and recognition is a basic and necessary function module. This can enable future in-depth studies on intelligent transportation while also lowering the likelihood of traffic accidents.