Traditional shadow detectors often identify all regions of static images or video sequences. This work presents the Referring Video Shadow Detection (RVSD), which is an innovative task that rejuvenates classic paradigm by facilitating segmentation particular shadows in videos based on descriptive natural language prompts. novel RVSD not only achieves arbitrary areas interest descriptions (flexibility) but also allows users to interact with visual content more directly and naturally using...

Abstract For bearing fault diagnosis problems in extremely noisy environments, this paper proposes an innovative universal adversarial training method. This method dynamically introduces noise into the data, adaptively optimizing model’s robustness. It applies to any neural network without incurring additional computational overhead reasoning process. Additionally, we introduce multi-scale channel attention (MSCAN). employs stacked convolutional kernels of varying sizes extract features at...

Six-dimensional (6-D) object pose estimation plays a critical role in robotic grasp, which performs extensive usage manufacturing. The current state-of-the-art techniques primarily depend on matching keypoints. Typically, these methods establish correspondence between 2-D keypoints an image and the corresponding ones 3-D model. And then they use PnP-RANSAC algorithm to determine 6-D of object. However, this approach is not end-to-end trainable may encounter difficulties when applied...

Accurately predicting pedestrian movements in complex environments is challenging due to social interactions, scene constraints, and pedestrians' multimodal behaviors. Sequential models like long short-term memory fail effectively integrate features make predicted trajectories comply with constraints disparate feature modalities of trajectory. Though existing convolution neural network (CNN) can extract features, they are ineffective mapping these into for pedestrians struggle model...

Abstract In recent years, deep learning has made significant strides and found extensive applications in tool wear prediction. However, most methods based on rely large-scale datasets for training do not consider the method under variable working conditions. To address these limitations, this paper proposes a novel pre-trained fine-tuning model graph-labeling graph neural network (GGPT) specifically designed to train small-scale datasets. The GGPT comprises two essential components: feature...

Abstract This paper introduces a comprehensive performance evaluation algorithm explicitly designed for secondary equipment in substations, specifically targeting the relay protection system. In contrast to current systems, this novel method navigates complex internal interconnections and mechanisms inherent within system equipment. Such complications have previously impeded accuracy breadth of evaluations, thereby limiting degree precision innovation attainable substations. The proposed...

Surface reconstruction using neural networks has proven effective in reconstructing dense 3D surfaces through image-based rendering. Nevertheless, current methods are challenging when dealing with the intricate details of large-scale scenes. The high-fidelity performance rendering is constrained by view sparsity and structural complexity such In this paper, we present Res-NeuS, a method combining ResNet-50 surface for reconstruction. Specifically, appearance embeddings: used to extract depth...

Reducing the load swing angle is very important for a crane system, and an input shaping technique PID algorithm are used to design composite control scheme crane. Through simulations, zero vibration derivative shaper selected from three ones open-loop suppression system. Then, double feedback structure based on displacement angle. Finally, designed by combining system with In terms of suppression, framework better than or algorithms, its advantages effectiveness verified simulations.

The improvement of endurance mileage and ride comfort for electric vehicle are both typical issues. Considering energy recovery vibration mitigation, a vertical-longitudinal comprehensive control method the with in-wheel switched reluctance motors (IWSRMs) is proposed in this paper. Firstly, after analyzing generating characteristics IWSRM, dynamic system model established, which includes motor drive system, braking IWSRM-suspension integrated system. Then, force distribution controller...

To address the problem of low trajectory prediction accuracy in aerial combat due to factors such as high degree freedom fighter aircraft flight, diverse states, and uncertain air situations, a algorithm based on an attention mechanism combined with graph convolutional network encoder-decoder structure is proposed. First, time domain features extracted by long short-term memory are input into obtain its spatial features. Secondly, used enhance network's ability learn antagonistic...

Carrier transport is the fundamental factor determining macroscopic transient characteristics of gallium arsenide photoconductive semiconductor switch (GaAs PCSS), with pulse width as a visible manifestation. In this letter, influence capacitance on GaAs PCSS investigated at an optical excitation 136nJ. The results indicate that electrical gradually transforms from compressed high amplitude and narrow to typical nonlinear lock-on waveform increases. Furthermore, carrier dynamics evolution...

In view of the problems Ant Colony Optimization (ACO), such as too slow convergence, low search efficiency, and easy to fall into local optimum, an improved ACO based on Self-Organizing Map (SOM) neural network competition mechanism is proposed. The SOM used adjust pheromone distribution path. initial pheromones path are distributed unevenly by parameter normalization improve convergence speed ACO. order solve problem ACO, update threshold set. algorithm applied planning. simulations carried...

At present, the vast majority of fire rescue still relies on equipment assisted manual rescue, which has problem high labor cost and danger, making artificial intelligence extinguishing main direction future rescue. This design is aimed at needs intelligent robots, using STC89C52 microcontroller minimum system, sensor detection technology, control camera recognition combined with raspberry pie positioning etc., to achieve automatic flame robots. Based results, signals are given orient...

Abstract Vision is the most important way for human beings to obtain information. Under constant evolution of electronic imaging technology, visual images are extensively applied production and life people. The analysis image information can achieve intelligent control complete specific tasks in industrial production. For example, logistics parcel sorting, traditional manual sorting slow, inefficient costly. system, machine vision was used information, depth learning algorithm locate...

Abstract The rapid development of substations has increased the demand for accurate and fast fault prediction systems. In order to achieve localization autonomous decision‐making modules types in substations, article proposes a algorithm based on improved ant colony optimization (IACO) back propagation neural network (BPNN). data substation secondary equipment training testing BPNN model is actual operating substation, which can significantly improve reliability results. addition, IACO used...

This paper introduces the domestic and foreign research status quo of flexible-link manipulator, to turn single manipulator in plane as object.To establish dynamic model system by using method modal analysis Lagrange's equation.This also has studied LQR control theory applied it system; The simulation is established MATLAB Simulink, compared effect vibration system.It pointed out importance choosing appropriate weighting matrix for LQR.By theoretical study analysis, obtained a satisfactory...