990 results about "Goal recognition" patented technology

The invention discloses an unmanned aerial vehicle cruising method and system based on a deep learning artificial intelligence image recognition algorithm. According to the method, an unmanned aerial vehicle carrying an improved YOLOv7-SwinT target recognition model collects real-time image data of an inspection area, and the model fuses a single-stage target detection architecture of YOLOv7 and a visual feature extraction network of Swin Transform. Progressive target detection is realized by adopting a three-level recognition architecture, wherein the progressive target detection comprises primary anomaly detection based on lightweight CNN, intermediate accurate positioning in combination with an attention mechanism and advanced target classification of multi-sensor data fusion. And the system combines the electric quantity of the unmanned aerial vehicle, the environmental condition and the task priority according to the identification result, generates a dynamic inspection path through an adaptive path planning algorithm, and realizes multi-vehicle collaborative operation by using an intelligent task allocation algorithm. In the inspection process, sensor data are processed in real time through edge computing equipment, and charging scheduling is optimized by adopting an intelligent energy management system. The target recognition precision and the cruising efficiency of the unmanned aerial vehicle in a complex environment are remarkably improved, and the method is suitable for application scenes such as electric power inspection and security monitoring.

The invention relates to the field of image processing, and discloses an object monitoring method and device based on multi-source data and a storage medium, and the method comprises the steps: synchronously collecting multi-channel video streams, environment parameters and equipment position information, and carrying out the time alignment and data association processing, and forming associated data; performing feature matching on the multiple paths of video streams, fusing position information and environment parameters, and establishing a mapping relation between video feature points and a unified space coordinate system; splicing the multiple paths of video streams in real time according to the mapping relation to generate a panoramic video stream; performing dynamic scene analysis based on the panoramic video stream and the environmental parameters, and identifying a target type and a state to obtain an analysis result; and generating an equipment regulation and control strategy based on the analysis result, and generating and issuing a regulation and control instruction for adjusting the working parameters of the multi-view image acquisition device based on the equipment regulation and control strategy. According to the invention, the splicing precision, the real-time performance and the target identification accuracy of panoramic monitoring can be improved, and the dynamic adaptive regulation and control of the equipment can be realized.

The invention discloses a target detection method and system based on a millimeter wave radar. The target detection method and system are used for realizing accurate detection, positioning and dynamic and static recognition of multiple targets under a complex background. According to the method, a distance-Doppler spectrogram is generated through the technical means of sliding window construction, spectral analysis, clutter suppression and the like, and candidate target points are detected by adopting an SO-CFAR algorithm. Then, determining a target position through high-resolution direction estimation and coordinate transformation, performing spatial clustering in combination with a density-based DBSCAN algorithm, and extracting a target geometric center and a bounding box; in the aspect of target tracking, Kalman filtering is used for predicting and updating the position and speed of the target, and a beam forming technology is used for enhancing a target signal, so that the target recognition stability is improved. And finally, the system performs robust dynamic and static state recognition on the target through a dynamic and static judgment module, so that high precision and robustness of the target detection process are ensured. The method can effectively cope with static background interference and dynamic target changes, and is suitable for target detection and tracking in a complex environment.

The invention discloses a visible light, infrared and IQ signal fusion individual identification method based on cross-modal cross attention, and belongs to the technical field of artificial intelligence and multi-modal image processing. Aiming at the problems of insufficient multi-modal heterogeneous feature fusion capability, unbalanced modal semantic expression and unstable classification precision in the prior art, a visible light image, an infrared image and an original IQ signal are acquired, and after preprocessing, a convolutional neural network is combined with a space attention module to extract image features; using a convolutional hybrid network to extract signal spectrum features; three groups of modal pairs are constructed by adopting a cross-modal bidirectional cross attention mechanism to carry out bidirectional semantic interaction and feature fusion; and finally, inputting the fusion features into a classifier to obtain an identification result. According to the method, deep semantic fusion can be realized, modal quality changes can be dynamically adapted, and the precision and robustness of target recognition in a complex environment are improved.

The invention discloses a dynamic scene incremental reconstruction and rendering method based on 3DGS, and belongs to the field of specific computer models, and the method comprises the steps: constructing a 3DGS model at an initial moment based on an original image set; obtaining any visual angle image at the moment t in the dynamic scene, and determining a first updating area through semantic segmentation and target recognition; generating an increment updating region based on the luminosity error, the local similarity and the global semantic feature; generating a second update region, modeling in the second update region, and minimizing region reconstruction errors to generate an optimized Gaussian point set; and fusing and optimizing the real-time model at the previous moment to obtain a real-time model Gt at the moment t for real-time rendering. According to the method, geometric and texture information of a scene can be effectively coded, accurate detection and local increment updating of a dynamic region are supported, Gaussian point parameters are optimized to improve the continuity and visual quality of a model, low-delay real-time rendering is realized, and the efficiency and quality of three-dimensional scene processing in a dynamic environment are effectively improved.

The invention discloses an unmanned aerial vehicle and bird target identification method and system based on multiple modes, and belongs to the technical field of computer vision and target identification. The method comprises the steps of collecting an RGB image, an infrared image and a continuous frame image of a monitoring area, extracting visual features, thermal radiation features and motion features after preprocessing, fusing multi-modal features by adopting an adaptive weight fusion strategy, detecting a potential target through a multi-scale feature fusion detection head, and obtaining a target detection result. The multi-mode classification module judges the target category and triggers early warning; the system comprises a data acquisition module, a preprocessing module, a feature extraction and identification module, a decision and early warning module and a database management module. Through multi-modal fusion, a self-adaptive weight strategy, multi-scale detection and multi-branch classification, the method can effectively overcome the limitation of a single modal, improves the recognition accuracy, the small target detection capability and the environmental adaptability in a complex environment, and meets the requirements of real-time recognition and early warning.

The invention provides a universal visual perception method for an open world based on language guidance, and belongs to the crossing field of computer vision and natural language processing, and the method comprises the steps: generating an initial multi-modal fusion representation, and obtaining the initial multi-modal fusion representation through the collection of a video frame sequence and a language instruction through processing and fusion; then determining a target candidate region, and carrying out matching, screening and optimization based on semantic keywords and visual region features; then generating a target identifier, and allocating a unique identifier to the high-confidence region; a continuous tracking trajectory sequence is formed, and a bounding box is updated and the trajectory is smoothed in combination with an algorithm; when the target disappears, the state vector of the target is temporarily stored, and the tracking identifier is recovered when a similar region appears; optimizing the tracking sequence, and adjusting a bounding box to generate an optimized sequence; and finally, outputting target motion trail, position and state information. Through multi-modal fusion, an optimization algorithm and a recovery mechanism, the open world target identification and tracking effect is effectively improved, and the practical value is high.

A modular open system architecture for common intelligence picture generation is disclosed. The system receives intelligence requirements through a multimodal artificial intelligence system and calculates collection feasibility across multiple intelligence sources based on physical and temporal conditions. The system develops integrated collection plans through the containerized analytics workbench using containerized analytics modules and processes intelligence through GPU-accelerated deep learning models for automated target recognition. Satellite collection is orchestrated through satellite data acquisition optimization platform by evaluating weather conditions, orbital parameters, and sensor capabilities, while space domain awareness is maintained through space domain awareness system for real-time collection asset management. Multi-source intelligence data is fused through multi-source intelligence fusion system to populate a common intelligence picture. The system implements automated workflows for intelligence analysis and dissemination while maintaining security controls, with the containerized analytics workbench providing pattern of life analysis and dynamic exploitation through containerized microservices.

The invention provides a text processing method and device, electronic equipment, a storage medium and a program product, and relates to the technical fields of artificial intelligence, natural language processing, large language models, automatic driving, intelligent traffic and the like. The method comprises the following steps: determining an initial voice recognition text and phoneme data of a target voice, and generating target prompt information based on the initial voice recognition text and the phoneme data; obtaining a target recognition text through a large language model based on the target prompt information; the large language model can correct the initial speech recognition text based on the phoneme data; as the accuracy of the phoneme data is far greater than the accuracy of the initial speech recognition text, more and more effective information can be provided for reference for the processing process of the large language model by combining the phoneme data, the large language model is helped to obtain a correct text in the processing process, and then the accuracy of text processing is improved.

The invention discloses an electric shovel surrounding multi-target identification method and device fusing multi-modal data, equipment and a medium, and relates to the field of target detection. Information data is preprocessed, a semantic mask is determined based on a semantic segmentation network, and cross-modal interaction from an image to a point cloud is carried out; according to the motion increment and the multi-modal data key features, determining a motion track set and a detection frame of each detection target around the electric shovel, and adopting an adaptive weight optimization mechanism to optimize the detection frame and a corresponding motion track based on a total loss function; an LM algorithm is adopted to carry out iterative optimization by taking a minimum total loss function of a multi-task loss system as a target, and secondary optimization detection frame track information and a detection frame after secondary optimization are determined, so that a multi-modal space-time depth fusion detector can be attached to appearance detection frames corresponding to detection targets around the electric shovel; and perception detection identification of a three-dimensional target is realized. According to the invention, multi-modal space-time deep fusion can be realized, and the detection precision is improved.

The invention belongs to the technical field of computer vision, and discloses a target identification tracking method based on a self-supervision mechanism, and the method comprises the steps: enhancing a self-supervision pre-training module through causality, constructing a causal sample pair through unlabeled video data, learning universal features through combining with comparison loss, and achieving the high-precision tracking without large-scale manual labeling. A multi-modal feature fusion and dynamic calibration mechanism further reduces dependence on annotated data, is especially suitable for industrial inspection, field monitoring and other scenes where data acquisition is difficult, significantly reduces time and labor costs in a data preparation stage, and broadens the application range of the technology in resource limited scenes; a causal reasoning and physical constraint mechanism is introduced, a dynamic relation between targets is modeled through a space-time causal graph, unreasonable tracks are filtered in combination with a physical rule, and complex conditions such as shielding, rapid movement and extreme weather are effectively dealt with; the dynamic feature calibration module corrects feature drift in real time, and ensures stable model performance in long-term tracking.

The invention relates to sensing and tracking of intelligent driving in rain, snow, fog, night and large shielding scenes. In order to solve the problems of target invisibility, track interruption and misconnection caused by low visibility, a shielded dynamic target identification and tracking method based on deep learning is provided; according to the method, under a unified spatial index, sparse point cloud and road topology, passable and traverse areas, static shielding volume coding, ray marking visibility boundary and shielding entrance and exit are carried out; generating a motion voting field with consistent visibility through multi-agent situation reasoning, performing normal directional enhancement according to a passable boundary, and extracting a risk corridor; voxelization is carried out on the multi-frame point cloud in the corridor, a dynamic sparse voxel map is constructed, a voting field is used for gating cross-frame edge connection, occupation changes and infinitesimal displacement are aggregated, and three-dimensional verification candidates, limited state estimation and a time continuous track are obtained; outputting the target and the corridor to which the target belongs, and forming a potential conflict zone according to the intersection of the target and the own vehicle path; the reproduction rate and the advance are improved, and cross-lane misconnection is reduced.

The invention discloses a multi-modal data fusion method in electric power overhaul, which comprises the following steps of: firstly, respectively acquiring long-sequence video data and audio data in an electric power overhaul process, and constructing a window-level multi-modal data fusion model; secondly, performing time window division and coarse alignment on the long-sequence video data and audio data by the window-level multi-modal data fusion model, generating a frame-level quality score, and performing fine alignment on multi-modal characteristic data in a time window by adopting a segmented time sequence alignment and compensation algorithm; and calculating a window-level quality score, a window-level weak supervision label and a confidence coefficient, and finally performing multi-modal fusion according to the multi-modal feature data after fine alignment and the window-level weak supervision label to generate a window-level semantic representation as a model output. According to the method, accurate time alignment of video and audio data and consistent fusion of multi-modal semantic features are realized, and more reliable alignment data and semantic input are provided for subsequent target recognition.

The invention relates to a humanoid robot indoor action planning and action control method and system and a robot. The method comprises the steps that an environment image is collected and analyzed; the large model is adopted to convert original text execution input by a user into a structured command; constructing an open set scene graph; after the target description text is received, if the place is a strange place, the no-graph task planning module generates a humanoid robot action instruction according to the target description text and the environment image; if the scene is a long-term activity place, the graph task planning module adopts LLM for processing according to the open set scene graph and the task instruction, and outputs an action instruction sequence which is executed in sequence; according to the robot action instruction or the action instruction sequence, a corresponding action control strategy is called, and physical movement of the humanoid robot is controlled. According to the method, autonomous action planning and control of the humanoid robot in an indoor environment are realized by utilizing strong zero sample target recognition and semantic understanding reasoning capabilities based on image and language basic models.

The invention relates to the technical field of intelligent sorting control. The multi-modal sensing sorting control method comprises the steps of obtaining multi-modal original sensing information, processing the multi-modal original sensing information, extracting category information and three-dimensional space pose information of a to-be-sorted target, generating a target recognition result, and recognizing the to-be-sorted target on the basis of the target recognition result. The method comprises the following steps: executing a path planning operation to obtain sorting path information containing a grabbing point location and a movement track, obtaining preset target clamping pose information, and generating a clamping control instruction based on the sorting path information and the preset target clamping pose information. The method has the effect of improving the sorting efficiency in a complex operation scene.

The invention relates to the technical field of computer vision, and particularly discloses a road intersection detection method used in an optical remote sensing image, a road intersection detection model is constructed, the model takes Swin Transform as a backbone network to carry out feature extraction, deep fusion of local details and global context information is realized, and the detection accuracy is improved. A lightweight encoder structure and a structure perception matching mechanism are combined, so that the spatial feature modeling efficiency is improved, and meanwhile, the structure expression capability of target recognition is enhanced; and meanwhile, a structure semantic vector is introduced to participate in a joint matching loss optimization process, so that the target representation precision and the structure consistency are further improved, and the model is more stable in representation in a complex intersection detection scene. Compared with a traditional CNN detection method and a standard DETR model, the road intersection detection model constructed by the method is remarkably improved in the aspects of detection precision, structure distinguishing capability and robustness.

The invention discloses a multi-modal task automatic driving perception method, device and equipment, and relates to the technical field of automatic driving systems, and the method comprises the steps: obtaining multi-modal data from a plurality of sensors, the multi-modal data comprising image data, point cloud data and radar data; performing time sequence hybrid coding on the multi-modal data to generate fusion time sequence features; based on the fusion time sequence feature, performing weighted fusion on each modal feature to obtain a multi-modal fusion feature; and inputting the multi-modal fusion feature into an end-to-end multi-task processing network to execute a sensing task to obtain a sensing result. According to the scheme, the target recognition accuracy is improved, and the overall generalization ability and scene adaptability of the system are remarkably enhanced.

Some embodiments automatically identify a target of a robotic process automation (RPA) activity (e.g., a button to click, an input field to fill out) according to a semantic similarity between a design-time label of the target and a label of a target candidate selected from a runtime instance of the target UI. Semantic similarity herein denotes likeness of meaning, as opposed to wording. Some embodiments employ a language model (LM) to quantify semantic similarity.

The invention relates to the technical field of image processing, computer vision and emergency rescue, in particular to a rescue target recognition method and system based on image recognition, and the method comprises the steps: obtaining a target image shot by an unmanned plane, carrying out the preprocessing of the image, and extracting a current contour of a combustion region; judging whether the contour shape is a regular shape or not by adopting multi-scale differential geometric analysis, and accordingly, adopting different unburned area identification strategies: for the regular shape, analyzing and judging an unburned area based on a current contour and a convex hull; the method comprises the following steps: performing shape regularity analysis on an irregular shape and an obtained inscribed polygon or a circumscribed rectangle, analyzing safety by constructing a distance field function, predicting and evaluating time safety margin in combination with fire spreading, comprehensively considering factors such as area and shape of a region, determining an optimal rescue target position, and generating a rescue instruction containing priority. The target identification accuracy in a complex fire environment is improved, the false alarm rate and the missing report rate are reduced, and the near-real-time processing speed is realized.

The invention discloses a multi-level and multi-mode target identification and dynamic tracking method based on an unmanned aerial vehicle, and relates to the technical field of target identification and tracking, and the method comprises the steps: obtaining target information in the tracking process of the unmanned aerial vehicle, processing visual features and semantic features through a double-flow heterogeneous network, and generating a fusion feature map; recognizing a search object in the fused feature map, locking a tracking target based on a target evaluation and motion prediction result, and starting real-time dynamic tracking of the unmanned aerial vehicle; and periodically calculating the coordinate position of the tracking target relative to the unmanned aerial vehicle, adaptively adjusting the calculation frequency through regional warning monitoring, and continuously tracking the tracking target. According to the method, different levels of target identification are established, the object range is defined through the target contour, the target to be tracked is determined through the color, the vehicle mark, the animal hair color, the dressing and vehicle color and the like, effective combination of target identification and a dynamic tracking technology is realized, and automatic searching and automatic tracking of the target needing to be searched can be completed.

A system for nuanced target recognition, comprising one or more processors coupled with memory, the one or more processors may be configured to detect, using a second model, an object based on a sequence of images, determine, using the second model, a class of the object for one or more of the images of the sequence of images, based on the images and an output of a first model, wherein the output comprises class definitions associated with a plurality of objects, generate a classification of the object based on the determined classes for the one or more images, and present the object and the classification on a display coupled with the one or more processors.

The invention relates to the technical field of risk early warning, and provides an operation risk early warning method, device and system for substation operation and maintenance personnel and a medium. According to the implementation scheme, the method comprises the steps of training a behavior recognition model based on multi-modal historical data to obtain a target recognition model; based on the current operation data of the operation and maintenance personnel, performing risk identification on the current operation behavior represented by the current operation data through a target identification model to obtain an identification result; if the identification result is that the current operation behavior belongs to the dangerous behavior, triggering a multi-dimensional risk assessment mechanism to perform risk assessment on the current operation behavior to obtain a risk assessment result; and generating risk early warning information based on the risk assessment result. According to the embodiment of the invention, the intelligent early warning of the operation risk of the operation and maintenance personnel of the transformer substation can be realized, and the safety guarantee level of the operation and maintenance personnel during operation is improved.

The invention discloses a dynamic environment and multi-event collaborative awareness method based on a multi-modal neural network. According to the method, multi-modal data is obtained through a multi-source sensing device, and a multi-modal data sequence under a unified reference coordinate is formed through time synchronization and space registration. And inputting the data sequence into a multi-modal deep neural network, and adaptively correcting each modal feature weight according to a confidence coefficient change rate between time slices to obtain corrected feature information. And performing external target identification and internal event identification on the feature information to generate a candidate event set, establishing an event priority relationship, and outputting a collaborative event result including an event tag, a spatial position and a time range. And finally, the risk level is determined according to a collaborative event result, and a corresponding early warning or control instruction is generated, so that intelligent identification and collaborative decision-making of multiple events in a complex dynamic environment are realized, and the real-time performance and accuracy of the system in multi-source data fusion and safety monitoring are improved.

The invention relates to the technical field of sterilization equipment control, in particular to an intelligent control method and device for sterilization equipment and electronic equipment. The method comprises the following steps: deploying a sensing network in a disinfection space and executing environment calibration to generate sensing reference data; synchronously acquiring space environment data through a sensing network; constructing a target recognition model based on the sensing reference data and the space environment data so as to perform dynamic target recognition and contact judgment; determining environment state parameters according to dynamic target recognition and contact judgment results; executing risk partitioning according to the environment state parameters to determine a region type and a risk score; and establishing a killing strategy rule base and executing killing operation according to the area type and the risk score. According to the invention, a differential disinfection strategy is formulated and executed based on dynamic environment perception and partition recognition.

The invention discloses an underwater target identification and positioning method based on fusion of a physical model and deep learning, and belongs to the technical field of computer vision, and the method comprises the steps: carrying out the transmissivity estimation and physical restoration of a collected underwater monocular image according to an underwater light propagation model, and carrying out the image correction; key feature points are extracted, and high-quality matching point pairs are screened in combination with the joint similarity; further deriving a basic matrix through the high-quality matching point pairs meeting the epipolar geometric constraint relation, solving an essential matrix, and obtaining relative attitude parameters between the cameras in combination with weighted re-projection-LM optimization; and finally, refraction correction triangulation is carried out according to the Snell's law, pixel-level fusion is carried out after scale normalization and space alignment are carried out on the refraction correction triangulation and dense depth output by the MiDaS, three-dimensional space coordinates of the target are inverted, and high-precision recognition and positioning of the underwater target are achieved. The system is light in structure, efficient in calculation, suitable for being integrated on various underwater autonomous or remote control robot platforms and used for tasks such as target recognition, tracking and positioning.

The invention relates to the technical field of video behavior recognition, and discloses a video abnormal behavior recognition method fusing spatio-temporal characteristics. The method comprises the following steps: a video data modeling step: modeling according to a current video frame sequence and preset behavior characteristics to obtain a video behavior model; an experience pool forming step of dividing a plurality of experience layers according to historical identification result differences and record confidence by means of historical video abnormal behavior records to form a multi-layer experience pool; a strategy determination step of determining an intelligent identification strategy of each stage based on a video behavior model target behavior state, and screening a multilayer experience pool according to record confidence and a target matching degree to obtain an experience identification strategy; a strategy adjustment step: dynamically adjusting the two strategies by using a dual-channel mechanism to adapt to a real-time video environment, and determining a target identification strategy; and a behavior state acquisition step of analyzing the video frame sequence according to an identification instruction to acquire an actual behavior state, the identification instruction being generated based on the target identification strategy and the current video frame sequence.

The invention provides a multi-source traffic video data fusion management method and system. The method comprises the steps of collecting multi-source traffic video data and static metadata, performing target identification, performing association fusion based on the target identification data and the static metadata, determining an association video source and generating a target continuous spatial-temporal trajectory, performing event detection based on the target identification data and the static metadata, and performing event detection based on the target continuous spatial-temporal trajectory. Standardized semantic event information is generated based on the target continuous spatial-temporal trajectory, visual angle scoring is performed on each associated video source based on the standardized semantic event information, and occlusion scoring and image definition scoring are performed on each associated video source. And dynamic resource allocation is carried out in combination with the visual angle score, and a complete historical trajectory map of a target related to the event is output, so that collaborative perception of massive heterogeneous traffic monitoring videos, accurate event detection and optimal scheduling of system resources can be realized.

The invention discloses a water conservancy project safety monitoring intelligent edge computing terminal, system and method integrated with AI recognition. The terminal comprises an acquisition interface unit, a network communication unit, a data storage unit, IOT middleware and a central processing unit. The central processing unit comprises a data analysis module, an AI identification module and an alarm generation module; the data analysis module performs time sequence integrity, logic rationality and multi-source consistency verification on the preprocessed structured data; the AI identification module dynamically schedules NPU and GPU resources to run a lightweight AI video identification model to realize target all-weather detection and target identification; and when the data analysis module judges that the sensor monitors that the data is wrong or abnormal, the AI identification module is triggered to perform recheck verification, and the alarm generation module is started according to a verification judgment result. According to the invention, through multi-source data association verification and a hierarchical multi-mode early warning algorithm, more accurate anomaly identification and risk assessment of hydraulic engineering safety are realized.

The invention relates to the technical field of aerial target identification, and discloses an aerial dynamic target accurate identification method and system. The method comprises the following steps: acquiring multi-dimensional spectral data of a target through a multi-spectral imaging device, extracting spatial distribution characteristics, establishing a three-dimensional motion track model according to the spatial distribution characteristics, and calculating an instantaneous velocity vector; a motion deviation index set is generated in combination with preset reference motion parameters, abnormal behavior categories are divided through clustering analysis, and a key frame sequence is marked; reconstructing a local motion characteristic spectrum based on the key frame sequence, extracting morphological change parameters, and inputting the morphological change parameters into a pre-trained target recognition network to obtain a type recognition result and a confidence score; when the confidence coefficient is lower than a threshold value, activating a supplementary recognition process, acquiring high-resolution texture data, and fusing the high-resolution texture data with the initial recognition result to generate a final recognition tag; and updating the feature database and adjusting the network weight according to the final identification tag.

The invention discloses a traffic accident video evidence analysis system and method based on multi-modal deep learning, and relates to the technical field of deep learning, and the method comprises the steps: carrying out the target recognition and tracking through radar point cloud, extracting a target number component and a motion variance component, carrying out the calculation through normalization and linear combination, and obtaining a dynamic fusion weight of each modal; and quantization and adaptive adjustment of scene complexity are realized. Then extracting each modal feature vector in a preset time sequence window, and performing weighted fusion based on the weight to form a unified feature vector; and inputting the uniform feature vector into a pre-trained accident classification model, and outputting whether an accident occurs or not and a type result. The method has the advantages that through complementary enhancement of multi-modal data, the accident identification accuracy in complex scenes such as low illumination and rain and fog is improved; adaptive adaptation to different scenes is realized through dynamic weight distribution; the whole scheme forms a closed loop, and has high robustness and practical value.