Search

Target detection technology in the field of machine vision plays a vital role in industrial production and manufacturing. In industrial production, productivity can be improved by accurate target detection. To implement this technology, many enterprises must manually clean and label a huge dataset. Meanwhile, it is a huge challenge for enterprises to obtain the dataset because of enterprise data privacy and security constraints. This paper proposes a method for rapidly generating synthetic samples based on digital twins to address this challenge. First, the virtual environment is utilized to replicate the real detecting environment, generating a variety of sample photos. The three-dimensional coordinates of the target object are then extracted in the virtual scene. Subsequently, an annotation method is designed for synthetic samples obtained from the virtual scene, utilizing principles of three-dimensional coordinate transformation and perspective coordinate transformation. This method efficiently produces numerous labeled samples with diverse annotations. Ultimately, the model performs detection tasks in the actual world using the synthetic samples as training data. The experimental results show that the synthetic samples created by this method based on digital twins can substitute real samples and effectively identify target objects during actual detection tasks. This paper proposes a unique strategy for synthetic samples that reduces sample collection costs and privacy risks, thereby addressing the limitations of machine vision detection technology induced by sample limitations.

Traditional agricultural practices of hand-picking ripe tomatoes are labor-intensive and inefficient for large-scale harvesting. To address this, we propose an innovative approach using the YOLOv7 algorithm for ripe tomato detection, enabling robotic arms to perform the picking. However, the occlusion of tomatoes in the field often leads to unclear target features, causing false or missed detections. So it is worth studying and this paper proposes a tomato detection method based on improved YOLOv7. The novelty is shown below. First, a new structure called ReplkDext is redesigned to increase the receptive field. ReplkDext is introduced before the last layer of CBS in the backbone. Secondly, to overcome the problem of low FLOPS caused by frequent access to memory in traditional neural networks, the head structure of YOLOv7 is redesigned. By using FasterNet to optimize the structure between Concat and CBS in the head, FasterNet makes the model balance between running speed and detection accuracy. Finally, to improve the ability of convolution, ODConv is added after the last ELANN-2 structure in the Head layer. ODConv improves the feature extraction ability of small targets and obtains more feature information about ripe tomatoes. Experiments show that compared with YOLOv7, Map@.5 of Tomato-YOLOv7 has increased by 1.3%. The model is overall better than other models. The overall contribution of the Tomato-YOLO model is to provide important insights into agricultural product detection and provide a theoretical basis for automated tomato harvesting in orchards.

Recent developments in vehicle automation require simulations of human-robot interactions in the road traffic context, which can be achieved by computational models of human behavior such as game theory. Game theory provides a good insight into road user behavior by considering agents’ interdependencies. However, it is still unclear whether conventional game theory is suitable for modeling vehicle-pedestrian interactions at unsignalized locations or if more complex models like behavioral game theory are needed. Hence, we compared four game-theoretic models based on two different payoff formulations and two solving algorithms, to answer this question. Unlike the most previous studies that employed naturalistic datasets to test and validate such models, this study utilized a distributed simulation dataset to test and compare the models. The study was conducted by connecting a CAVE-based pedestrian simulator to a motion-based driving simulator to replicate the traffic scenarios for 32 pedestrian-driver pairs. The findings demonstrated that there is a high variability between participant pairs’ behaviors. Our proposed behavioral game-theoretic model outperformed other models in predicting the interaction outcome. This translates to a decrease by 70% and 67% in the root mean squared error (RMSE) when compared to the baseline model, for marked and unmarked crossings, respectively. The model can also predict which interaction will take the longest time to resolve. According to our results, road users cannot be expected to behave in line with the Nash equilibrium of conventional game theory that underscores the complexity of human behavior with implications for the testing and development of automated vehicles.

Regional cross-border electricity trading is popular in price and cooperation. It is a reliable path to alleviate the shortage of power supply during rapid economic growth. As an important export commodity, the pricing options for hydropower directly relate to the cooperation intention of business parties. This study demonstrates the price formation methods and internal relationships of cascade hydropower stations in CBET. To simulate the decision-making of a bilateral forward contract, a bi-level optimization model was constructed, and the objectives of each layer represent the interests of both parties. The BOM was linearized with Karush-Kuhn-Tucker conditions and strong duality, and finally solved using mixed-integer linear programming. The prices of CHSs in southwestern China were calculated under different pricing options and the results were absorbed in an actual transaction case. Consequently, the complementarity between different exporting countries was verified. The most suitable pricing option for CBET was determined, that is the unified electricity price based on the marginal cost, to be $481~\textrm {CNY/MW}\cdot \textrm {h}$ , which is 20% to 60% higher than the thermal and photovoltaic from other countries. Furthermore, national policy support and economic subsidies are key guarantees for the sustainable development of CBET, which can reduce the discount electricity price to below that of thermal generation by 30%.

In recent years, semantic segmentation based on deep convolutional neural networks has developed rapidly. However, it is still a challenge to balance the computing cost and segmentation performance for the current semantic segmentation methods. This paper proposes a lightweight real-time semantic segmentation model Balanced Sample Distribution Network (BSDNet), to solve this problem. In BSDNet, we introduce the Balanced Sample Distribution Module (BSDModule) to balance the sampling distribution of convolution and obtain features with a larger receptive field. To optimize the segmentation effect, we introduce a Shuffle Channel Attention Module (SCAModule) to enhance the interaction of channel features at the cost of a small number of parameters. BSDModule and SCAModule are lightweight and flexible and can adapt to different types of network structures. Extensive experiments on CityScapes and CamVid show that the proposed method can balance the computing cost and segmentation performance. Specifically, on the CityScapes with $512\times1024$ resolution, BSDNet-Xception39 achieves 68.3% MIoU and 84.6 FPS with only 1.2M parameters.

In order to achieve good connectivity after the cascading failure of a logistics network, this paper studies the controllability robustness of complex logistics network based on the nonlinear load-capacity (NLC) model. Firstly, the extended Baraba' si and Albert (BA) network is constructed as a complex logistics network for experiments, based on the power law distribution and the agglomeration and sprawl evolution mechanism. Secondly, the existence of the NLC relationship of the real logistics network is proved, and then the NLC model of complex logistics networks is proposed. Furthermore, a simulation analysis of the controllability robustness and influencing factors of the complex logistics network is carried out under four different cascading failure models. In those models, different scenarios of the NLC and the classical linear load-capacity (LLC) model with initial load (IL)/initial residual capacity (IRC) load-redistribution strategies are combined. The research results show that the main influencing factors of the cascading failure of complex logistics networks for the controllability robustness P' are the tolerance parameters β and γ. Moreover, the effect of γ on the load-capacity relationship under the NLC model is more significant than that of β. Among the four cascading failure models, the one based on the NLC model with IRC strategy is the optimal for controllability robustness. Based on the optimal model, the simulation considering the perspective of the logistics economy shows that the relationship among the network cost e, Pi and γ is as follows: under a fixed cost, the greater is γ, the stronger is Pi. Also, when 2

Inspired by the empirical dynamic characteristics of the load of real logistics network, we propose a dynamic cascading failure model against cascading failure, which is more suitable for complex logistics network by adding dynamic factors based on the nonlinear load-capacity model under initial residual capacity load-redistribution strategy. The simulation is executed on the complex logistics network model and the results show that the controllability robustness and economy after cascading failure based on the dynamic cascading failure model is feasible and effective. It can effectively reduce the logistics cost and enhance controllability robustness against cascading failure by adjusting the network cost e and capacity parameter γ, so as to balance the controllability robustness and economy for the complex logistics network.

To improve the precision and robustness of Unmanned Aerial Vehicle (UAV) integrated navigation systems, this paper presents an Interacting Multiple Model (IMM) navigation algorithm based on a Robust Cubature Kalman Filter (RCKF) with modified Zero Velocity Update (ZUPT) method assistance. This algorithm has a two-level fusion structure. At the bottom level, the Global Positioning System/Inertial Navigation System (GPS/INS) integrated navigation model and the Dynamic Zero Velocity Update/Inertial Navigation System (DZUPT/INS) integrated navigation model are established by modifying the Zero Velocity Update (ZUPT) method. Subsequently, the RCKF algorithm adopts a robust factor to weaken the influence of measurement outliers on the filter solution. At the top level, the estimation results of the GPS/INS integrated navigation model and the DZUPT/INS integrated navigation model are fused by the IMM algorithm. In addition to enhancing the robustness of filter estimation in the presence of measurement outliers, the proposed navigation algorithm also corrects navigation errors with ZUPT method assistance. Simulation and experimental analyses demonstrate the performance of the proposed navigation algorithm for UAVs.

We demonstrate a three-component fiber optic seismometer for seismic observation, with no electronics in the sensor probe, but relying exclusively on optical fiber in the form of an unbalanced Michelson interferometer, with state-of-the-art performance characteristics. The fiber optic seismometer system is installed in the seismic observation cave at the Earthquake Administration Bureau of Jilin Province, in Changchun, Jilin, China, and has been operating without interruption for more than a year. Two minor earthquakes of different magnitudes recorded by the three-component fiber optic seismometer are analyzed and discussed in detail from the perspectives of time domain and frequency domain signal quality and fidelity. In addition, the recording results of the fiber optic seismometer are compared with those of the traditional electrical-mechanical seismometer co-located at the observation site. Actual observational results of the seismic events show that the three-component fiber optic seismometer can record the seismic waves clearly, with comparable performance as the conventional seismometer.

Conventional ideal point model-based synthetic aperture radar (SAR) imaging techniques may lead to scattering estimation inaccurate and object-level information loss because they ignore the frequency, angle, and polarimetric dependence of canonical scatterers. These issues are especially severe when SAR measurement is contaminated with impulsive noise. To tackle the problem, in this paper, we propose a robust polarimetric SAR imaging method with attributed scattering priors in the framework of compressive sensing. In this scheme, the overcomplete dictionary is formed by taking advantage of the physically relevant attributed scattering center (ASC) model, which is capable to represent the electromagnetic behaviors of man-made assemblies. By using the robust data-fitting model and employing the joint sparsity of multiple polarization measurements, the problem of estimating ASC attributes can be converted into robust group sparse recovery via ℓ1-ℓ2,1 optimization. To solve this formulation efficiently, we develop an alternating direction method of multipliers framework via exploiting the proximity operator of complex ℓ1 norm and ℓ2 norm functions for complex-valued SAR images. Finally, the super-resolution SAR image is achieved via spectrum extrapolation based on the estimated ASC attributes. The proposed technique can enhance the SAR image features arising from the non-ideal scattering components of the target and gain robust performance to impulsive noise. The experimental results confirm the effectiveness and efficiency of the proposed algorithm.

The phishing email is one of the significant threats in the world today and has caused tremendous financial losses. Although the methods of confrontation are continually being updated, the results of those methods are not very satisfactory at present. Moreover, phishing emails are growing at an alarming rate in recent years. Therefore, more effective phishing detection technology is needed to curb the threat of phishing emails. In this paper, we first analyzed the email structure. Then, based on an improved recurrent convolutional neural networks (RCNN) model with multilevel vectors and attention mechanism, we proposed a new phishing email detection model named THEMIS, which is used to model emails at the email header, the email body, the character level, and the word level simultaneously. To evaluate the effectiveness of THEMIS, we use an unbalanced dataset that has realistic ratios of phishing and legitimate emails. The experimental results show that the overall accuracy of THEMIS reaches 99.848%. Meanwhile, the false positive rate (FPR) is 0.043%. High accuracy and low FPR ensure that the filter can identify phishing emails with high probability and filter out legitimate emails as little as possible. This promising result is superior to the existing detection methods and verifies the effectiveness of THEMIS in detecting phishing emails.

By exploiting the continuity structure of target scene, the problem of interrupted synthetic aperture radar (SAR) imaging for change detection is studied in this paper. Timeline constraints imposed on multi-function modern radars lead to gapped SAR data collections, which in turn results in corrupted image that degrades reliable coherent change detection (CCD). In this paper we extrapolate the missing data using the sparse Bayesian framework. In particular, the inherent clustered structures of the sparse target scene are characterized by structure-aware Bayesian priors. The variational Bayesian inference (VBI) is then utilized to estimate an approximated posterior of the sparse coefficients. Finally the CCD images are obtained by applying the coherence estimator to the resultant complex images. Based on the structural information in the imaging process, the devised method offers the advantages of preserving the weak scatterers and suppressing the artificial points with fewer measurements. Experimental results are presented to demonstrate the effectiveness and superiority of the proposed algorithm.

This paper presents a novel synthetic aperture radar (SAR) imaging and autofocus algorithm by exploiting the clustering property of the sparse scene in structured phase-noisy environments. The SAR imaging and autofocus problem are reformulated in a Bayesian framework where a clustered sparsity-inducing prior is imposed on the target scene and a correlated probabilistic model is assumed on phase error. In particular, a pattern coupled Gaussian distribution and a Markov random fields (MRF) model are assigned to the scattering coefficients and the support of the sparse scatterers, respectively, to enforce the underlying continuity of illuminated target scene. Furthermore, the multivariate Von Mises prior is incorporated to capture the spatial fluctuation structures of the phase errors along with the aperture positions. The posterior inference is then obtained by using the mean-field variational Bayesian expectation-maximization (VBEM) method. By taking advantage of the inherent target correlation characteristics and the uncertain structures introduced on phase errors, the devised method can result in enhanced imaging and autofocus performance. The experimental results are provided to demonstrate the superiority of the devised scheme in obtaining the concentrated target region over the other reported sparsity-driven SAR imaging and autofocus algorithms.

A photonics-based radar for high-resolution 3D microwave imaging is proposed and demonstrated, in which high range resolution is achieved by employing a wideband linearly frequency modulated signal, and high cross-range resolution in both azimuth and elevation is realized by rotating the target around two orthogonal axes to compose a 2D inverse synthetic aperture. The proposed radar performs photonics-based in-phase/quadrature up-conversion and de-chirp processing in the transmitter and receiver, respectively, which features a compact structure with low-sampling-rate electronics in the receiver. An experiment is carried out. The established radar works in the K-band, with bandwidth as large as 8 GHz. Captured echo signals at different rotation angles of the target are processed to a 3D image, in which the voxel values are scattering intensities at different spatial positions.

Real-time city distribution strategies are highly dependent on dynamic environments, requiring timely responses to real-time changes due to various dynamic events that take place in the distribution system. Considering the influence of four kinds of real-time information on vehicle routing and vehicle scheduling, including new requests arriving gradually, old requests being modified or canceled, traffic congestion and vehicle breakdowns, a dynamic vehicle routing model based on a dynamic pick-up and delivery problem considering multiple dynamic events in a real-world environment (DPDP-MDE) is established. A dynamic algorithm framework is designed to solve the problem, the tabu search (TS) algorithm and the adaptive large neighborhood search (ALNS) algorithm are adopted to improve the quality of the initial solution, and the dynamic insertion method is adopted to solve the synchronization problem of unfixed requests (that is, unaccepted customer requests and modified requests) and new requests. The experimental results show that the model and dynamic algorithm framework proposed in this paper can effectively solve the dynamic pick-up and delivery problem with time windows (DPDP-TW). At different scheduling time horizons T, the TS algorithm improves the initial solution by an average of 3.11% and the ALNS algorithm by an average of 9.98%. Under different degrees of urgency, compared to the ALNS algorithm, the quality of the solution produced by the TS algorithm is not high, but the computation time is very small and it is relatively stable. Under different request sizes, the TS algorithm can obtain optimization results in 60s under four request levels, which gives it a significant advantage over the ALNS algorithm.

Recovering missing values plays a significant role in time series tasks in practical applications. How to replace the missing data and build the dependency relations from the incomplete sample set is still a challenge. The previous research has found that residual network (ResNet) helps to form a deep network and cope with degradation problem by shortcut connection. Gated recurrent unit (GRU) can improve network model and reduce training parameters by update gate which takes the place of forgetting gate and output gate in long short-term memory (LSTM). Inspired by this finding, we observe that shortcut connection and mean of global revealed information can model the relationship among missing items, the previous and overall revealed information. Hence, we design an imputation network with decay factor for shortcut connection and mean of the global revealed information in GRU, called decay residual mean imputation GRU (DRMI-GRU). We introduce a decay residual mean unit (DRMU), which takes full advantage of the previous and global revealed information to model incomplete time series; and the decay factor is applied to balance the previous long-term dependencies and all non-missing values in the sample set. In addition, a mask unit is designed to check the missing data existing or not. An extensive body of empirical comparisons with other existing imputation algorithms over real-world data and public dataset with different ratio of missing data verifies the performance of our model.

Rail wear inspection is vitally important in the railway industry. Conventional methods mainly use manual or static measurements to detect rail wear, which are inefficient, imprecise, and unreliable. To improve the accuracy and efficiency of rail wear inspection, a dynamic detection system based on a revised fast global registration (RFGR) algorithm was employed. First, the framework for online detection of rail wear with multi-profile was put forward to reduce the influence of vibrations of individual sections. Second, the RFGR method was proposed by using a robust weight function to convert the non-convex registration model to a convex problem, and the Levenberg-Marquardt method was used to solve nonlinear least-squares systems robustly. Finally, the Hausdorff distance was introduced to visualize the distance between the wear profile and the reference profile after alignment. The experimental results demonstrated that the RFGR algorithm was more accurate, robust, and effective than iterative closest point (ICP), sparse ICP, Vi's sparse ICP, and the fast global registration algorithm. For actual wear detection, the proposed method was more efficient and robust for the online dynamic detection of rail wear when compared with the single-profile-section-based inspection method.