4250 results about "Edge device" patented technology

Provided is a road crack detection method, a medium and a product. The lightweight YOLO-MCS road crack detection model is constructed based on an improved yolov8 network. A method of improving the yolov8 network includes: replacing a feature extraction network backbone with a lightweight convolutional neural network MobileNetV3, and embedding a Coordinate Attention Mechanism (CA) module in the lightweight convolutional neural network MobileNetV3; adding a small target detection layer and a Squeeze and Excitation (SE) module at a neck end; and introducing a loss function of power Intersection over Union (IoU) into a head prediction structure. The road crack detection method overcomes the problem that the existing detection algorithm has low detection precision for small cracks and is difficult to be applied to edge devices with limited computing resources.

A system and method for adaptive edge-cloud data processing dynamically distributes computational tasks between edge devices and cloud infrastructure in response to changing conditions. The system continuously monitors resource availability, network parameters, and workload characteristics while predicting future conditions using hierarchical forecasting models. A multi-objective optimization approach determines optimal task distribution, balancing processing latency, energy consumption, bandwidth utilization, and result quality. The system implements a partitionable processing pipeline that enables seamless task migration through state synchronization protocols and checkpoint mechanisms. During migration, the system preserves processing continuity by establishing dependencies, creating execution checkpoints, and verifying successful state transfer. Performance metrics may be continuously collected and analyzed to improve future decision-making. The system maintains operational resilience during connectivity disruptions through local decision-making capabilities and eventual consistency protocols, making it suitable for diverse applications including industrial IoT, connected vehicles, healthcare wearables, and smart city infrastructure.

An AI-enhanced distributed system for neural network-based data compression leverages reinforcement learning optimization and privacy-preserving computation across edge and central computing devices to autonomously optimize efficiency and quality. The system includes a lightweight compression subsystem at edge devices that applies privacy-preserving preprocessing and partially compresses input data before securely transmitting it to central computing devices. A reinforcement learning agent continuously monitors system performance and automatically optimizes compression parameters, model selection, and task allocation based on multi-objective rewards. The central compression subsystem processes data using AI-optimized parameters and temporal modeling components. The system incorporates hardware detection capabilities that automatically select optimal compression models based on available processing resources and implements homomorphic encryption for computation on encrypted data while coordinating federated learning across distributed devices. This AI-enhanced distributed approach improves bandwidth efficiency, energy consumption, and adaptability while ensuring data privacy and security.

The invention relates to the technical field of special equipment intelligent monitoring and maintenance, in particular to a special equipment monitoring and maintenance method and system based on multi-dimensional data fusion, and the method comprises the steps: collecting and processing multi-source heterogeneous sensor data; constructing a dynamic digital twinborn model based on sensor data and a cross-domain term mapping rule; associating the sensor data with the digital twinborn model to establish a lightweight analysis model, performing health state simulation analysis on the edge equipment, and generating an analysis result; the data quality of the edge device is monitored through the analysis model, when the data quality is lower than a threshold value, a co-simulation process is started, a simulation analysis task is transferred to the cloud digital twin platform, and the edge device continues to conduct simulation analysis based on the analysis model; and constructing a simulation design decision rule base, and generating a maintenance suggestion. Through the multi-dimensional data fusion and edge cloud co-simulation architecture, the accuracy, real-time performance and reliability of monitoring and maintenance of the special equipment are improved.

The invention relates to the technical field of environmental sound event detection, in particular to an environmental sound event detection method based on multi-modal data fusion, which utilizes a microphone array, a camera and a sensor to synchronously acquire audio, video and environmental data to form a multi-modal data source. And the collected multi-modal data is preprocessed, so that the data quality and the accuracy of subsequent analysis are improved. And key features are extracted from the processed data, and the features are fused to form multi-modal features. And a deep learning model is constructed, and multi-modal features are used for training, so that a model capable of identifying an environmental sound event can be obtained. And performing optimization such as pruning, quantification and knowledge distillation on the trained model, and deploying the model on edge equipment to realize real-time preliminary feature extraction and event detection. The problems that an environment sound event detection method in the prior art is poor in robustness, low in real-time performance and limited in generalization ability are solved.

The invention provides a micro-grid fault diagnosis and dynamic recovery method based on deep reinforcement learning, and the method focuses on the multi-modal features of key nodes through a graph attention mechanism, extracts fault features through a multi-layer graph attention layer, and captures the spatial dependence relation of micro-grid nodes to recognize a fault propagation path. Meanwhile, spatial features and historical multi-modal data are fused with the help of a gating circulation unit, space-time joint feature representation is constructed, pre-fault symptom time sequence evolution is captured, intermittency and early fault detection capacity are enhanced, the multi-modal feature data fusion problem is solved, and high-precision fault diagnosis is achieved. A knowledge distillation technology is adopted to deploy a lightweight student model at edge equipment, millisecond-level emergency response is realized, fault diffusion is prevented, and meanwhile, the accuracy of diagnosis and repair strategies is guaranteed. The optimal repair strategy is generated at the cloud through the teacher model by using the global data, the system can adapt to the topological change of the micro-grid and novel faults, and the fault processing capability is continuously improved.

The invention discloses a power grid abnormal flow detection method based on multi-modal data fusion, and the method comprises the steps: collecting the multi-source heterogeneous data of a power grid through an edge calculation node, including the current waveform of an intelligent electric meter, the state variable of an SCADA system, network protocol metadata and an equipment log event; performing space-time alignment preprocessing on the original data, converting an unstructured log into a time sequence by adopting a sliding window mechanism, and eliminating sensor noise through an LSTM auto-encoder; constructing a multi-dimensional feature space which comprises a frequency domain feature, a spatial-temporal feature and a protocol feature, and obtaining a feature vector; the feature vectors are input into a hybrid detection model, the model is composed of an isolated forest algorithm, an improved CNN-LSTM classifier and an information entropy-based rule engine which are connected in parallel, and a dynamic weight fusion strategy is adopted to output an abnormal probability; and when the abnormal probability exceeds a dynamic threshold, triggering a multi-level response mechanism: sending a traffic shaping instruction to the edge device, generating a device fingerprint portrait on the cloud platform, and storing abnormal event features through a block chain.

The invention discloses a product detection method and system based on machine vision, and relates to the technical field of machine vision, and the method comprises the steps: initializing an ambient light sensor, evaluating an initial illumination condition, and dynamically adjusting a light source parameter to verify and lock an optimal setting; selecting a preset wavelength range to configure a spectral imaging system to collect image data according to the characteristics of the building material, performing primary processing and image synthesis to obtain a comprehensive spectral image, and performing feature extraction to obtain a high-quality image data set; preprocessing an image data set on edge equipment, deploying a deep learning model to carry out preliminary defect detection, identifying and marking defect types, and extracting local and frequency domain features; and refining defect types by using a classification algorithm according to the preliminary detection result, integrating a final classification result and image data, and transmitting to a cloud to complete detection. According to the invention, high-precision and high-efficiency product defect detection is realized.

The invention discloses a data real-time analysis and decision-making system based on an edge intelligent Internet of Things, belongs to the technical field of Internet of Things, and aims to solve the technical problems of insufficient real-time performance, high delay, large bandwidth occupation and high dependence on a central server in data processing of Internet of Things equipment. Comprising an edge device layer, an edge computing node layer, a cloud edge collaboration module and a central server layer, each edge computing node is used for taking data after local analysis processing as input, performing prediction analysis through a local model, making a decision through a decision generation mechanism based on an analysis result, and performing cloud edge collaboration. Generating decision optimization through an adaptive decision optimization algorithm, and performing data exchange with the central server layer through a cloud edge collaboration module; and the central server layer is used for carrying out model updating based on the real situation and the analysis result corresponding to the analyzed and processed data uploaded by each edge computing node, and returning the updated model parameters to the edge computing node layer.

The invention discloses a multi-agent non-cooperative game driven high-concurrency task reasoning method and system, and the method comprises the steps: decomposing a heterogeneous task into standardized task units, and carrying out the quantitative modeling of the calculation resource demands of the task; constructing a task auction mechanism of the multi-agent non-cooperative game model, performing task allocation by adopting an anti-strategic auction rule and a fragmentation asynchronous protocol, and constraining resource declaration behaviors of the agents through a credit pledge mechanism; the real-time resource utilization rate of the system is monitored, the task priority is dynamically adjusted in combination with the task preemption historical state, a resource soft preemption strategy is triggered, and a compensation queue is established; a self-adaptive model splitting strategy is adopted to distribute reasoning tasks to end side equipment and edge computing nodes, cooperative execution is achieved through data compression and transmission, and task rescheduling is conducted according to feedback of an execution result; according to the method, the task execution efficiency, the resource utilization rate and the system stability can be remarkably improved in a high-concurrency scene.

The invention provides an electric power multi-modal data perception and digital twinborn simulation cooperative interaction method and system, and relates to the field of electric power equipment management and maintenance, and the method comprises the steps: carrying out the electric power multi-modal data perception of data collected by edge equipment, and obtaining the state characteristics of single equipment; constructing a digital twinborn model based on the state features of all devices in the to-be-monitored power system; constructing a power equipment graph knowledge base by using a digital twinborn model, and extracting global state features of the equipment from the power equipment graph knowledge base through a graph neural network so as to construct a global state feature time sequence of the equipment; inputting the global state feature time sequence into the trained time sequence prediction model, and predicting the equipment state at the future moment; according to the multi-modal data, parameters of the digital twin model are adjusted in real time, environmental changes are flexibly adapted, time sequence modeling and fault trend analysis functions are combined, the running state of the equipment is predicted, and the comprehensiveness and foresight of equipment monitoring are further improved.

The invention relates to a power edge cloud cooperative management method based on a swan OS, and the method comprises the following steps: S1, deploying intelligent edge nodes, setting an embedded AI acceleration unit, operating the swan OS on all the intelligent edge nodes, and achieving the seamless cooperation of equipment through the distributed characteristics of the swan OS; and S2, realizing self-organizing connection of edge devices by adopting a Mesh network technology, designing a direct communication path between the edge devices, and realizing rapid processing of local problems. S3, aiming at the real-time requirement of a power system, optimizing an MQTT / CoAP protocol stack, improving the efficiency and reliability of message transmission, and allocating special network resources for different types of power applications by utilizing the 5G network slice characteristics, S4, constructing a security architecture based on a zero-trust principle, enhancing the equipment access control and data transmission security, and S5, establishing a security architecture based on a zero-trust principle. And identity authentication and data integrity protection are performed by using a block chain technology, so that data tampering and counterfeiting are prevented. According to the method, the local distributed decision-making capability and the global cooperation capability are effectively improved.

A system and method for temporal acceleration encoding in Lorentzian latent space enables real-time event forecasting within navigable spatiotemporal media. The system encodes media data into compact Lorentzian latent patches using variational autoencoders and organizes them within a multi-dimensional hyperspace spanning spatial, temporal, orientation, scale, and spectral coordinates. Temporal acceleration encoding computes velocity and acceleration vectors along geodesic trajectories, extracting event signatures through multi-scale aggregation over sliding windows. An acceleration-indexed memory stores dynamic descriptors with composite keys comprising hyperspace coordinates and motion characteristics. Event forecasting retrieves similar historical patterns and conditions a forecast head to produce event probabilities and time-to-event estimates with uncertainty calibration. The system streams forecast metadata to edge devices for real-time prediction and adaptive navigation, supporting applications in surveillance, autonomous systems, predictive media exploration, and anomaly detection where both temporal forecasting and multidimensional navigation capabilities are essential.

The invention discloses an intelligent security and protection monitoring method and system based on image recognition. The method comprises the following steps: S1, acquiring multi-modal data; s2, performing preprocessing to generate complete target information; s3, performing real-time detection on the generated complete target information, and performing primary processing on a conventional scene; s4, based on an improved visual SLAM technology, obtaining three-dimensional information of multiple cameras; s5, when the edge device detects a potential abnormal condition, performing abnormal behavior analysis on the data by using a Swin Transform model, and optimizing a detection result in combination with the three-dimensional information of the multiple cameras; s6, performing adaptive learning on historical data by using a depth deterministic strategy gradient algorithm, and dynamically adjusting a threshold value of abnormal behavior detection; and S7, after an abnormal condition is detected, processing data through differential privacy protection and an end-to-end encryption technology, automatically controlling security and protection equipment, and giving a real-time alarm. According to the invention, image recognition and deep learning technologies are fully utilized, and accurate real-time monitoring in a complex scene is realized.

The invention provides a multi-feature fusion rumor detection method, system and device based on knowledge distillation, and mainly solves the problems that an existing model is high in calculation overhead, insufficient in feature fusion and insufficient in emotion utilization. The method comprises the steps of firstly obtaining multi-dimensional data such as social media original texts and comments; extracting deep semantic representation by using a pre-training model, and analyzing comment emotion features in combination with a hybrid neural network; then, features such as semantics, emotions, emoticons and populations are input into a hierarchical gating interactive fusion network (GIFN), and weights are dynamically adjusted to achieve effective fusion of multi-granularity features; in order to reduce complexity, a knowledge distillation framework is designed: a deep GIFN is used as a teacher network to generate a soft label, and a lightweight student network (LSTM) is guided to perform training. According to the trained student model, the parameter quantity is remarkably reduced, meanwhile, good detection performance is kept, the student model can be conveniently deployed in an actual content auditing system or edge equipment, and social content rumors can be efficiently recognized and judged.

The invention discloses an intelligent advertisement putting processing method and system based on user data, and relates to the technical field of data processing, and the method comprises the steps: putting an advertisement in an advertisement pool based on an edge device, collecting the video information during the advertisement playing period through the edge device, analyzing the stored video information, extracting the user feature information, and storing the user feature information in the advertisement pool; the user feature information is integrated into user feature vectors, the user feature vectors stored in the edge devices are uploaded to the cloud platform, the cloud platform analyzes the user feature vectors uploaded by the edge devices to generate user portrait information, and the user portrait information is displayed on the basis of the user portrait information generated by the cloud platform. And putting corresponding advertisements to the user groups with similar user portrait information, collecting feedback data, and dynamically optimizing the advertisements put by the edge devices and the user groups based on the feedback data. The method has the advantages that the data transmission delay and the cloud computing pressure are reduced through the cloud edge collaboration method.

The invention relates to the technical field of dynamic migration, in particular to an edge computing management system supporting dynamic migration of an AI model. The system comprises a state mining module used for identifying available edge computing devices; performing multi-frequency information monitoring on the available edge computing devices, and performing device comprehensive state mining to generate edge device comprehensive state features; the edge topology module is used for carrying out node position positioning one by one based on edge equipment comprehensive state characteristics and carrying out instant state mapping modeling so as to construct an edge computing node topology model; the task flow trend module is used for acquiring an AI model task running log; and performing task queue analysis on the AI model task running log, and then performing model task traffic trend analysis to obtain AI model task traffic trend data. According to the invention, edge computing management with quick response and high processing efficiency is realized.

The invention discloses an improved YOLOv11-based lightweight aerial image target detection method, which comprises the following steps of: firstly, acquiring an aerial image data set and converting the aerial image data set into a YOLO format, and then constructing an aerial image detection model which takes a YOLOv11 model as a reference network and comprises a backbone network, a neck network and a head network; in the backbone network of the aerial image detection model, a Conv module in the backbone network of the YOLOv11 model is replaced by a DBR Down module, and an MSFLB module is added in the neck network; and finally, performing iterative training on the constructed aerial image detection model to obtain a trained aerial image detection model, and detecting a to-be-detected target in the aerial image by using the trained aerial image detection model to obtain a detection result. The aerial image detection model constructed by the invention can effectively improve the precision of aerial image target detection under the condition that edge equipment resources are limited.

The invention discloses a bearing residual life prediction method based on a multi-teacher element weight knowledge distillation network, and the method comprises the following steps: (1) carrying out the normalization, noise reduction and segmentation of a multi-dimensional vibration time sequence signal collected by an original sensor, and dividing the signal into a training set and a verification set; (2) inputting the processed training set data into a multi-teacher element weight knowledge distillation network, training teacher models, and fixing parameters of the three teacher models after training is completed; (3) performing knowledge distillation training on the student model by using an adaptive meta-weight strategy gradient learning algorithm; and (4) the student model after distillation training is used for bearing residual life prediction. According to the method, the number of network model parameters is small, the calculation complexity is low, the prediction precision is high, and the method can be actually deployed on edge equipment.

An artificial intelligence based Offensive Cybersecurity Appliance or OCA for launching a cyber-offensive countermeasure aimed at effectively mitigating sophisticated cyber threats such as real-time Ransomware as a Service (RaaS) directives and emergent killware instigated by malicious-threat actors or cyber attackers. The system covertly uses a vast array of penetrative counter cyber attacks, based on an Advanced Persistent Threat or APT model, to circumvent encountered information security controls instituted by security-based components of a cyber attacker's Local Area Network or LAN. Intrinsically, the penetrative counter cyber attacks are constituted of multiple blended artificial intelligence based cyber attacks. The OCA subsequently uses these blended artificial intelligence based cyber attacks to render the hardware (including the cyber attacker's computer system used to instigate the cyber attack, discovered peripheral smart devices, and network-edge device such as a modem-router) interconnected to the cyber attacker's LAN inoperable by destroying the electronic components associated with that hardware.

The invention discloses an AI server intelligent data analysis method based on edge-cloud collaboration, and relates to the technical field of edge-cloud computing collaboration, edge equipment is deployed at an edge end, original data of each task is collected in real time, the data is preprocessed, task difficulty features are extracted, and a task difficulty feature sequence is formed; and based on the task difficulty feature sequence and a pre-trained AI data analysis model, the task difficulty of the collected task data is evaluated in combination with the current resource condition, and the feasibility of task execution at the edge end is judged. According to the method, equipment is deployed at the edge end and data is preprocessed, so that part of computing tasks are effectively unloaded to the edge end, the computing burden of a cloud AI server is remarkably reduced, the edge end can independently complete simple tasks, complex tasks are uploaded to the cloud according to needs, excessive centralized use of cloud resources is avoided, and the computing efficiency is improved. Therefore, the energy consumption and the operation cost of the cloud server are reduced, and the energy efficiency ratio of the whole system is improved.

The Internet can be configured to provide communications to a large number of Internet-of-Things (IoT) devices. Devices can be designed to address the need for network layers, from central servers, through gateways, down to edge devices, to grow unhindered, to discover and make accessible connected resources, and to support the ability to hide and compartmentalize connected resources. Network protocols can be part of the fabric supporting human accessible services that operate regardless of location, time, or space. Innovations can include service delivery and associated infrastructure, such as hardware and software. Services may be provided in accordance with specified Quality of Service (QoS) terms. The use of IoT devices and networks can be included in a heterogeneous network of connectivity including wired and wireless technologies.

The invention relates to the related field of artificial intelligence (AI) and edge computing, in particular to an AI algorithm process and service all-in-one machine based on edge computing, and efficient operation and dynamic adaptability of a system are achieved through cooperative work and intelligent resource management of edge equipment and a cloud end. By deploying the real-time resource monitoring module on the edge device, the computing resource use condition of the device can be monitored in real time, resource allocation is dynamically adjusted according to the current task requirement, and the working efficiency and the energy utilization rate of the device are improved. And the cloud end performs regular optimization and issues the updated model to the edge device, so that dynamic updating and optimization of the model are realized, and the system is ensured to maintain efficient and stable operation in a complex and changeable environment. Through preliminary reasoning of the edge device and auxiliary processing of the cloud end, the data transmission amount and the cloud computing burden are reduced, and meanwhile the robustness, maintainability and expandability of the system are enhanced.

The invention relates to the field of building intellectualization, in particular to a digital twinborn mixed cloud edge collaborative intelligent house building group operation and maintenance intelligent system, which establishes a digital twinborn scene database by collecting building basic information, equipment operation data and environmental parameters, synchronizes the digital twinborn scene database to edge equipment, and uses BIM, GIS, Internet of Things, 5G and AI technologies to establish a digital twinborn scene database, so as to realize the intelligent operation and maintenance of a building group. A virtual-real combined digital intelligent building scene is constructed, real-time synchronization of a virtual scene and a physical environment is realized through AR / VR equipment, and an operation and maintenance module comprises multi-source heterogeneous data fusion, edge intelligent analysis decision, adaptive model training and iterative optimization, a predictive maintenance algorithm of virtual-real mapping and a multi-level collaborative decision and autonomous scheduling mechanism. And the monitoring module monitors the state and operation condition of the edge equipment, provides data service and supports visualization of management decisions, and the system effectively improves the intelligence and digitization level of operation and maintenance of the building group.

An intelligent anomaly detection system (100) for IoT networks using edge computing and deep learning, comprising: (a) a data collection and pre-processing module configured to collect data from a variety of heterogeneous IoT devices and perform pre-processing operations, including normalization, denoising, and feature extraction; b) an edge intelligence and model deployment module configured to deploy deep learning models optimized for edge devices to perform local data analysis; (c) an anomaly detection and classification module configured to identify abnormal behavior in the processed data using deep neural networks and to classify the anomalies into predefined threat categories; (d) an adaptive learning and model update module configured to update the deployed models through incremental or federated learning mechanisms without transmitting raw data to a centralised server; (e) an alert and response management module configured to generate alerts and execute predefined mitigation actions based on the type and severity of the detected anomalies; f) and a system monitoring and visualization module configured to display real-time insights, alerts and system analysis through a user interface, g) the system operates in a decentralised manner using edge computing to achieve scalable and privacy-preserving anomaly detection in IoT environments in real time.

The invention relates to the field of edge computing, and discloses an intelligent recognition algorithm model edge-end fusion type deployment method, which comprises the following steps: collecting original data from different sensors, acquiring data from various sensors in real time by edge equipment, and carrying out primary processing and screening on the data; the edge device operates a lightweight intelligent recognition algorithm, processes the data in real time and makes a preliminary judgment; an intelligent task scheduling system is designed, and the processing requirement of each task is evaluated in real time; when the edge equipment cannot independently complete a calculation task, data is transmitted to the cloud, and the cloud processes a complex intelligent identification task by using powerful calculation resources to generate a high-level reasoning result; and the cloud transmits the optimized model or processing result back to the edge device, and the edge device performs adaptive adjustment according to the feedback of the cloud to optimize the local processing capability. The method has the advantages of improving the response speed and the computing power.

The invention discloses a classroom teaching quality auxiliary evaluation method, device and system based on cloud edge collaboration, and the method comprises the steps: obtaining video image data, collected by a camera, of students in a classroom in real time, and transmitting the video image data to an edge device; and in the edge device, performing student classroom behavior identification and tracking on the received video image data by using a pre-trained student classroom behavior identification model, and fusing the identification results of adjacent multiple frames with the identification result of the current frame in the tracking process to obtain the final identification result of the current frame. Evaluating the classroom teaching quality according to the student classroom behavior recognition result, wherein the student classroom behavior recognition model is obtained by training based on a multi-label target detection algorithm and by adopting a knowledge distillation structure in advance; and sending a student classroom behavior identification result and an evaluation result to a cloud video server for storage. The method has the advantages of being simple in implementation method, low in cost, high in real-time performance and precision, high in robustness and the like.

An intelligent power monitoring management system based on Internet of Things belongs to the technical field of power supervision and comprises a data acquisition module, a communication transmission module, a data processing and analysis module, an edge calculation module, an intelligent control module, an early warning management module and an authority control module. According to the invention, by deploying various types of sensors, real-time acquisition and dynamic monitoring of key parameters of the power system are realized. The system can obtain and analyze data such as voltage, current and power in time, ensures the accuracy and high efficiency of operation of the power system, and effectively avoids energy consumption waste and equipment faults. The edge computing technology is adopted, preprocessing and preliminary analysis are carried out at a data acquisition end, the data processing pressure of a central server is relieved, the bandwidth requirement for data transmission is lowered, and the response speed is increased. Even if the network is interrupted, the edge device can realize short-term data caching, and data continuity is guaranteed. And performing anomaly detection on the power system, identifying potential risks and giving an alarm.

A method for processing a query for data stored in a distributed database includes receiving, at an edge device, the query for data stored in the distributed database from a query device. The method includes causing, by the edge device, the query to be stored on a dynamic ledger maintained by the distributed database. The method includes detecting, by the edge device, that summary data has been stored on the dynamic ledger. The method includes generating, by the edge device, an approximate response to the query based on the summary data stored on the dynamic ledger. The method includes transmitting, to the query device, the approximate response.

The invention discloses an image classification method and system based on CNN-Transform staged fusion, and the method specifically comprises the steps: inputting an input image into a lightweight CNN backbone network after preprocessing, and extracting the local features of a shallow layer, a middle layer and a deep layer; non-overlapping image blocks are divided on the middle-layer feature map, learnable position codes are embedded, and global semantic features are generated through linear attention Transform; a dynamic gating fusion module is designed, channel attention and space attention branches are combined to generate adaptive weights, and local and global features are subjected to weighted fusion; the CNN shallow detail features and Transform deep semantic features are fused through cross-level jump connection, and the classification robustness is improved. The system supports model quantization, operator fusion and edge device deployment, and the classification precision is remarkably improved while the calculation complexity is reduced.