146 results about "Image perception" patented technology

The invention discloses a complex water area garbage intelligent detection and collection method based on multi-modal perception, and the method comprises the steps: obtaining image perception data and semantic environment data of a complex water area; the method comprises the following steps: performing preprocessing, performing feature extraction and semantic coding, and then performing multi-modal feature alignment fusion, feature extraction and multi-task detection to obtain a garbage detection result, thereby effectively realizing deep fusion of semantic priori and visual features, and improving the accuracy and robustness of garbage recognition in a complex water area environment; then task importance evaluation is carried out to construct a garbage disposal priority queue, a path dynamic planning strategy oriented to energy consumption constraint and task priority collaborative optimization is constructed to carry out path generation, navigation energy consumption is reduced on the premise that task benefits are considered, and garbage detection and collection tasks in a complex water area are achieved; the system adapts to various intelligent devices such as unmanned ships, unmanned aerial vehicles and shore-based platforms, and has high autonomy, environmental adaptability and engineering deployability.

The invention relates to the technical field of rainstorm early warning, in particular to a deep learning-based rainstorm terminal early warning method and system, which introduces an image perception and semantic recognition mechanism to carry out consistency judgment on an overturning state of a rain gauge and image features such as a rain curtain and a water mark so as to realize effective elimination of unnatural trigger factors (such as manual water pouring and the like). And the reliability of rainfall sensing data is improved. A double-branch deep learning model capable of being deployed locally is adopted, multi-source data such as rainfall, temperature and humidity and topographic factors are fused, feature extraction and recurrence period prediction are performed based on a cross attention mechanism, and a threshold correction coefficient is dynamically generated in combination with historical events, so that self-adaptive adjustment of an early warning standard is realized, and misinformation and missing report risks are effectively reduced. The system supports low-power-consumption local operation, early warning decision is executed, and timeliness and accuracy of local early warning response are effectively improved.

The invention relates to the technical field of robot control, and discloses a robot motion control method, device, equipment and medium, and the method comprises the steps: constructing a simulation environment and configuring a robot model, obtaining a depth image of an area in front of the robot model through a depth sensor, carrying out the semantic segmentation of the depth image, generating a semantic mask of a passable area, and carrying out the recognition of the passable area. And generating three-dimensional terrain point cloud data, extracting a semantic feature vector of the point cloud data, inputting the semantic feature vector and the body state information of the robot model into the motion control strategy model to generate a joint control instruction, and driving the robot model to move according to the joint control instruction. According to the method, depth image perception, semantic segmentation, three-dimensional point cloud reconstruction and feature extraction are fused, robot body state information is combined, a self-adaptive joint control instruction is generated, dynamic perception, decision making and control of robot motion in various complex terrains can be achieved, and the autonomous motion energy of the robot in an unstructured environment is effectively improved.

The present invention relates to a compaction monitoring device and method that integrates 3D scanning and image perception. The device comprises a sensing component comprised of a 3D road surface texture scanning module, a camera, a vibration damping module, and a locator; a data analysis component, which performs image positioning, image stitching, time synchronization, and calculation of a compaction index based on texture features; and a control strategy component, which controls the number of compaction attempts based on the surface distribution of the compaction index. Compared to existing technologies, this invention introduces a new information source and evaluation basis for road compaction, enabling dynamic, real-time monitoring and feedback control of compaction levels, improving the quality of road construction projects and enabling compaction monitoring based on 3D scanning and image perception.

The invention provides a bridge structure health diagnosis agent construction method based on a thinking chain and a domain tool set. According to the method, a language model with a causal reasoning capability is introduced, and a complete task chain process is constructed by combining target detection, segmentation and geometric analysis models for image processing, so that automation and intelligence from image perception to structure diagnosis are realized. The method supports multi-image input and multi-task cooperative processing, has good generalization ability and interpretability, and can effectively improve the efficiency and accuracy of structural health assessment.

The invention relates to an insulator image segmentation method based on an infrared enhancement image of an unmanned aerial vehicle, and relates to the technical field of image processing, and the method comprises the steps: collecting an infrared image sequence through an infrared enhancement camera, obtaining a standard infrared image sequence, and carrying out the state recognition through an image perception prior engine, outputting a predicted insulator attention heat map and a predicted boundary confidence map; optimizing and adjusting the initial threshold segmentation algorithm and the initial region growing algorithm, and obtaining an adaptive threshold segmentation algorithm and an adaptive region growing algorithm; and carrying out image segmentation on the standard infrared image sequence to output an insulator image sequence, and carrying out early warning judgment on the insulator image sequence based on an adaptive early warning mechanism. The method solves the problems that a traditional insulator image segmentation method cannot effectively deal with the problems of low infrared image temperature contrast ratio, large noise interference and complex background, so that the insulator segmentation is easy to cause mistaken segmentation and missing segmentation, and the high-precision requirement of fault detection is difficult to meet.

The invention discloses a digital media content element accurate screening method based on artificial intelligence image recognition, and relates to the technical field of digital media, and the method comprises the steps: building a distributed capture network to form a dynamic content pool, and building a metadata index database; calling a multi-modal image perception engine to generate a double-layer characteristic spectrum containing dominant and recessive elements; constructing a distributed recognition model cluster based on federated learning; converting the user demand into a screening parameter set and generating a decision tree; screening and secondarily verifying an output result through a double-path matching mechanism; and constructing a reinforcement learning reward function based on user behaviors, and driving the model and the decision tree to co-evolve. Multi-source heterogeneous content full-dimension analysis is achieved, the recognition comprehensiveness and depth are improved, knowledge barriers and privacy risks are solved, the screening accuracy and flexibility are improved, the system is endowed with the continuous optimization capacity, and the method is suitable for efficient and accurate digital media content screening scenes.

The invention provides an electromagnetic adsorption type drilling actuator capable of achieving automatic feeding and an automatic drilling method, and relates to the technical field of intelligent machining. The electromagnetic adsorption type drilling actuator comprises a drilling machine body, a connecting support, a feeding system, an electromagnetic adsorption device, an image sensing unit and a control system; the connecting support is slidably connected with the drilling machine body and fixedly connected with the tail end of the mechanical arm. The feeding system is fixedly arranged on the connecting bracket; a moving part of the feeding system is fixedly connected with the drilling machine body; the electromagnetic adsorption device is fixedly arranged on the connecting support, and the adsorption face of the electromagnetic adsorption device faces the feeding direction of the drilling machine body. The image sensing unit is fixedly arranged on the drilling machine body; and the control system is electrically connected with the drilling machine body, the feeding system, the electromagnetic adsorption device and the image sensing unit. A visual module is integrated on the aspect of the device, an executing mechanism with the real-time environment sensing capacity is constructed, and cooperative control of precise feeding transmission and self-adaptive adsorption is achieved through modular design.

The present invention includes systems and methods for optimizing perception of image data. Color vision impairment affects a user's ability to distinguish between certain colors, reducing the impact of color images. Mapping color image data to an expanded color gamut increases contrast between colors and the available number of distinct colors in a set of image data, which improves the ability of a user with color vision impairment to see differences between colors. The present invention includes systems and methods for determining a type of color vision impairment in order to provide customized optimization of image data for improved color perception.

The invention provides an active train sensing system and method based on thunder and vision fusion. Comprising an acquisition module used for acquiring multi-path image data containing different focal length images and three-dimensional point cloud data corresponding to the front of a train; the time synchronization module is used for generating synchronous image data and synchronous three-dimensional point cloud data; the coordinate transformation module is used for establishing a spatial transformation relation between the three-dimensional point cloud coordinate system and the image coordinate system; the image sensing module is used for inputting the synchronous image data and the three-dimensional point cloud data into the target segmentation network to obtain track coordinates; the point cloud sensing module is used for obtaining three-dimensional point cloud data in front of the train; and the thunder-vision fusion module is used for distributing an accurate center representative coordinate for each obstacle, generating structured obstacle information and outputting the structured obstacle information to a vehicle-mounted train control system. The multi-source sensing information utilization rate can be improved, the track obstacle recognition and three-dimensional positioning precision is improved, and the system sensing stability in a complex scene is enhanced.

The invention discloses an image generation and understanding unification method and system based on joint diffusion modeling, and relates to the technical field of image generation and understanding. According to the method, image generation and understanding tasks are unified through joint diffusion modeling, models do not need to be designed for the generation and understanding tasks respectively, and the efficiency is improved. Through improved DINOv2, Segmenter and DETR models, classification feature class cluster polymerizability, boundary detail segmentation and small target feature expression detection are enhanced. A random role allocation mechanism and a mask full attention mechanism flexibly process multi-domain information, domain invariant position coding assists cross-domain alignment, and the modeling precision is improved. Through optimization training, the model supports joint generation, controllable generation and image perception tasks at the same time, the performance is superior to that of an existing unified model, and even exceeds that of a proprietary model in tasks such as edge detection.

The invention discloses an extensible reconstruction method for a super-large scene based on 3D (three-dimensional) Gaussian Splitting. The balance of the number of sub-block images is ensured through image perception self-adaptive blocking; a low-contribution view is removed through SSIM pruning; a priori depth is obtained based on virtual stereo rendering, a confidence mask is generated through multi-view verification, boundary consistency loss is constructed, and Gaussian densification is driven; and finally, carrying out TSDF fusion to obtain a seamless grid. Experiments show that reconstruction is completed in a 2.7 km scene within 6 h, MatrixCity-Aerial F1 reaches 0.784 and is improved by 41% compared with City GaussianV2, precision, efficiency and boundary consistency are considered, and the method is suitable for being applied to oversized scenes such as digital twin cities and automatic driving.

The invention provides a multi-source heterogeneous traffic light data fusion optimization method based on vehicle and road cloud, and the method comprises the following steps: S01, traffic light data collection: carrying out the collection of the traffic light data in three modes, the method comprises the following steps: S02, carrying out image perception and identification on traffic lights through a camera, obtaining traffic light data of an intersection through V2X short-distance communication, and obtaining traffic light data through cloud traffic light data service, S02, carrying out data cleaning, S03, carrying out multi-source data fusion, and S04, carrying out HMI data display. More than 99.9% of service availability is realized; the average delay is controlled within 100ms; and city-level scale deployment is supported. Under the extreme condition, if the communication delay fluctuation is large, the phenomenon of transient display delay or second skipping may occur, but the system can still ensure the basic service continuity through a data compensation mechanism.

The embodiment of the invention provides an image processing method, electronic equipment and a computer readable storage medium. The method relates to the technical field of artificial intelligence, and comprises the steps: carrying out the down-sampling of a target image, and obtaining image information; performing information extraction on the target image to obtain semantic information; coding is carried out based on the image information and the semantic information, and a composite code stream is generated; the composite code stream is transmitted to a server side for processing, an image processing result of the target image is obtained, and the server side is used for executing an image processing task for the target image. According to the technical scheme provided by the embodiment of the invention, by combining image semantics, the requirement of image transmission bandwidth is reduced, the data volume of semantic information is far smaller than that of pixel-level information, the transmission data volume can be greatly reduced, the image perception effect under an extremely low bit rate is realized, the intelligent perception analysis capability of an image understanding task is better supported, and the image understanding efficiency is improved. According to the invention, the technical problem that the image perception capability is insufficient under the condition that the bandwidth is limited in the prior art is solved.

The invention relates to a visual language detection method for detecting insects and pests based on open vocabularies, and the method comprises the steps: taking a to-be-detected image and the text description of the to-be-detected image as input, and extracting visual features and text features; and constructing a multi-modal feature fusion module, mapping the visual features and the text features to a unified semantic space, generating language-perceived visual features and image-perceived text features, and aligning the language-perceived visual features and the image-perceived text features in the unified semantic space by using a comparative learning mechanism. The system is designed for effectively detecting pests in the real world, and the key requirement of an adaptive pest recognition system is met; the multi-modal feature fusion module effectively integrates a visual language mode, and the performance of open vocabulary insect pest detection is remarkably improved by enhancing feature representation; regional cues show excellent generalization ability in an open vocabulary scene, thereby enabling more robust and flexible pest detection in different environments.

The invention provides an endoscope image sensing method and system based on self-supervised learning, and the method comprises the steps: constructing and training an endoscope image sensing model, obtaining a real-time endoscope image sequence pair, inputting the real-time endoscope image sequence pair into the endoscope image sensing model for sensing, outputting a predicted retracement endoscope image sequence, and outputting a predicted retracement endoscope image sequence. An image sequence matching mechanism based on deep learning is adopted, the predicted retracement endoscope image sequence and the real-time retracement endoscope image sequence are matched, difference evaluation and screening are carried out on the predicted retracement endoscope image sequence and the real-time retracement endoscope image sequence after matching, and a corresponding uncertainty evaluation result is generated; and performing interpretability prompting based on the uncertainty evaluation result, and optimizing a corresponding module of the endoscope image sensing system according to a feedback result of the interpretability prompting, thereby realizing efficient and accurate identification of the abnormal region in the endoscope image.

A system and a method for medical image processing are provided. The method includes: obtaining a first medical image; obtaining a trained image perception restoration model; the trained image perception restoration model includes an image quality perception sub-model and an image restoration sub-model; and inputting the first medical image into the trained image perception restoration model to obtain a second medical image. The image quality perception sub-model is configured to determine either or both of a first quality evaluation value of the first medical image and a second quality evaluation value of the second medical image, the image restoration sub-model is configured to determine the second medical image, and the quality of the second medical image is higher than that of the first medical image.

The invention discloses an underwater image enhancement method and related device based on adaptive image attribute learning, and the method comprises the steps: obtaining to-be-enhanced underwater image data, inputting the to-be-enhanced underwater image data into an image attribute complementary network, outputting image contrast feature information and image chroma feature information corresponding to the underwater image data to be enhanced; inputting the image contrast feature information and the image chroma feature information corresponding to the underwater image data to be enhanced into an adaptive image perception balance network for image fusion processing, and outputting fused image feature information; and inputting the fused image feature data into an image optimization network for image optimization processing, and outputting enhanced underwater image data formed after optimization processing. According to the embodiment of the invention, color cast and blurring can be effectively removed, the image definition can be effectively improved, and underwater image segmentation and key point detection tasks are facilitated.

The embodiment of the invention provides a multi-modal problem solving method and a multi-modal problem solving system.The multi-modal problem solving method comprises the steps that an input image is obtained, and the input image comprises a problem text and a problem graph; analyzing the question text to obtain a question analysis result and a target graphic element; performing vector graphic marking on the target graphic element in the topic graphic to obtain a vector graphic marking result; and based on a vector graphic marking result, reasoning a question analysis result to obtain an answer text. Attention on image details is improved in a vector graphic marking mode, image perception precision is improved, a vector graphic marking result can serve as an intermediate representation in a processing process, interpretability of a problem solving process and understanding of a verification method on the image details are enhanced, and a problem analysis result is reasoned based on the vector graphic marking result, so that the accuracy of image perception is improved. Direct perception of possible errors can be avoided, and the problem solving accuracy is improved.

The application provides a panoramic image splicing system and method for weak light enhancement and color temperature correction. The system mainly comprises a multi-thread control module for overall controlling the execution sequence of each module; a video reading and image correction module responsible for reading four camera images and performing image correction operations; a parameter initialization module for processing images, performing key point detection and matching, and calculating the rotation matrix between images and estimating camera parameters; an image projection module for projecting each image onto a cylindrical surface according to the rotation matrix and camera parameters; an optimal seam line finding module for calculating the optimal seam line of each overlapping area; an image fusion module for fusing images according to the optimal seam line and a multi-band fusion method to obtain a panoramic image; and a panoramic image display module responsible for displaying the panoramic image. The system can realize efficient and accurate panoramic image splicing and effectively handle weak light and color temperature problems to improve image perception.

The invention discloses an electric power inspection unmanned aerial vehicle with multi-mode autonomous ability, which is provided with a sensing ability framework based on a multi-mode sensing model, an end-to-end control network based on a deep neural network architecture and a task execution architecture, and the unmanned aerial vehicle collects body state information and inspection task environment images through a sensor; the perception capability framework understands the information of the collected inspection task environment image and generates a target 3D position; the end-to-end control network maps the body state information and the target 3D position to an action space of the unmanned aerial vehicle so as to control the unmanned aerial vehicle to reach the target position; and the task execution framework is combined with a sensing capability framework and an end-to-end control network to automatically complete an inspection task. The method has strong generalization ability in a complex and changeable electric power inspection environment, tasks can be executed in a highly automatic manner, the inspection difficulty and manpower and material resource consumption are reduced, and the inspection safety and reliability in a dangerous and complex environment are greatly improved.

The invention discloses an autonomous representation processing method, system and device and a storage medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: receiving a representation demand input by a user; in combination with a large language model and a preset thinking chain reasoning mechanism, analyzing and disassembling the representation demand to obtain a task process; executing imaging and positioning operation according to the task process, and obtaining imaging data; performing target object instance extraction and attribute calculation on the imaging data to generate an instance-level structured object; characterization conclusions and / or decision information or physical execution instructions are formed based on the instance-level structured object. According to the method, task autonomous planning is realized by using a large model, and automatic positioning, screening, measurement and statistics are supported by parallel visual analysis, so that the labor cost is remarkably reduced; the individual recognition precision in a complex scene is improved through dynamic process disassembly; an executable instruction is generated based on pixel-physical mapping, an image is opened to perceive a physical operation closed loop, and the problem of cross-device collaboration is effectively solved.

The application provides an intelligent image processing system and method for a car based on a perception algorithm model, and the method comprises the following steps: S1. spatio-temporal reference double anchoring and dynamic intrinsic extrinsic parameter calibration of a vehicle-mounted image acquisition node; S2. multi-node image heterogeneous domain normalization and adaptive preprocessing based on a self-adaptive kernel regression non-local mean denoising algorithm; S3. hierarchical image feature extraction and semantic anchoring based on a graph neural network dynamic feature interaction network; S4. cross-node and cross-frame feature mutual checking and pseudo-feature elimination; S5. full-scene semantic completion and dynamic target trajectory prediction based on a variational autoencoder trajectory prediction model; S6. dynamic lightweight adaptation and algorithm power adaptive scheduling of the perception algorithm model; and S7. risk scene grading identification and image targeted enhancement output based on semantics and trajectories. The application provides stable, accurate and efficient vehicle-mounted image perception support for intelligent driving of a car, and improves the safety and adaptability of environmental perception of intelligent driving.

The invention discloses an image perception method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining the image features of each visual angle image in a multi-visual angle image, and enabling the multi-visual angle image to be synchronously collected by a multi-visual angle image collection device on a carrier; determining a first conversion relation between an image coordinate system corresponding to each view angle image and a target coordinate system; based on the first conversion relation corresponding to each view angle image, projecting the image feature of each view angle image to a target coordinate system to obtain the projection feature of each view angle image in the target coordinate system; fusing the projection features of the view angle images in the target coordinate system to obtain fused features; and based on the fusion features, determining the perception result of the multi-view image, thereby improving the perception precision on the premise of satisfying the requirements of increasing the perception range and reducing the calculation amount.

The utility model relates to railway vehicle detection technical field, concretely relates to a kind of EMU high-voltage cable terminal acoustic image detection device, including acoustic image acquisition unit and acoustic image analysis unit, both constitute data connection signal transmission, wherein the acoustic image acquisition unit includes microphone array, camera, the microphone array is used for the acoustic perception of high-voltage cable, the camera is used for the image perception of high-voltage cable;The microphone array includes the MEMS type sensor combination of several array elements, and the arrangement of several array elements MEMS type sensor uses the irregular sparse array based on multi-arm spiral array.The utility model has the advantages that: realize EMU high-voltage cable terminal partial discharge abnormal sound non-contact detection, accurately master EMU high-voltage cable use condition and life cycle, scientific guidance roof high-voltage cable overhaul and replacement, can effectively guarantee EMU operation safety and stability, substantially reduce EMU operation and maintenance cost expenditure.

The application relates to an insulator image segmentation method based on an unmanned aerial vehicle infrared enhanced image, and relates to the technical field of image processing, and comprises the following steps: acquiring an infrared image sequence through an infrared enhanced camera to obtain a standard infrared image sequence, performing state recognition by using an image perception prior engine, and outputting a predicted insulator attention heat map and a predicted boundary confidence map; optimizing and adjusting an initial threshold segmentation algorithm and an initial region growing algorithm to obtain an adaptive threshold segmentation algorithm and an adaptive region growing algorithm; performing image segmentation on the standard infrared image sequence to output an insulator image sequence, and performing early warning judgment on the insulator image sequence based on an adaptive early warning mechanism. The application solves the problem that the traditional insulator image segmentation method cannot effectively deal with the problems of low temperature contrast, large noise interference and complex background of the infrared image, so that the insulator segmentation is prone to false segmentation and missed segmentation, and the high-precision requirement of fault detection cannot be met.

This invention discloses a method for scene classification of remote sensing images with a small sample size, belonging to the field of image perception. The method first divides the remote sensing image scene classification dataset into training, validation, and test sets; then, it constructs scenario sets based on these datasets; next, it constructs a deep nearest neighbor neural network model based on an attention mechanism; then, it trains and validates the attention-based deep nearest neighbor neural network model based on the scenario sets; finally, it tests the trained attention-based deep nearest neighbor neural network model based on the scenario sets. This invention addresses the problem of complex scene backgrounds in remote sensing image scene classification by proposing a deep neural network model based on an attention mechanism. It introduces a scenario training method, uses deep local descriptors for feature representation, and introduces an attention mechanism to obtain scene category-related attention maps. The deep local descriptors are divided into scene-related and scene-independent parts. A metric module calculates the similarity between the local descriptors of the query image and the category, and a weighted sum is performed using the attention map, thus solving the problem of scene classification in small samples of remote sensing images.

The application relates to a high-definition image explainable classification method and device, computer equipment and a storage medium. The method simulates the process of image perception, feature extraction, inductive reasoning and learning of the human brain. The method is based on the latent space of a pre-trained StyleGAN, a conversion network is designed to simulate the feature extraction of the visual cortex of the brain, effectively converting high-dimensional latent encoding into low-dimensional and decoupled classification features, and the problems of rule explosion and calculation collapse are alleviated; a fuzzy neural network is used to simulate the reasoning function of the parietal lobe and the prefrontal cortex of the brain, and an improved method of the fuzzy neural network structure suitable for the high-definition image explainable classification model is designed, so that the explainable classification of the low-dimensional features is realized; the fuzzy rules used for classification are visually displayed and analyzed, and a feature visualization method is designed based on the pre-trained StyleGAN generator, so that the explainability is further improved.

The application provides a terminal model adaptive updating method for edge video inference, which is improved through an existing edge-end cooperative system, adopts three core mechanisms of uncertainty-driven dynamic sampling, personalized pseudo-label generation of terminal device image perception and experience replay assisted online distillation training to construct an edge-end cooperative adaptive updating framework. Compared with the prior art, the scheme can effectively cope with the data drift problem caused by light changes, weather switching, view angle differences and the like in the edge scene, significantly reduces the terminal device upload bandwidth overhead, reduces the communication delay, suppresses the noise label interference, relieves the model catastrophic forgetting, fully adapts to the differences of large-scale heterogeneous terminal devices in computing power, data distribution and scene working conditions, realizes long-term stable inference with low delay, high precision and strong robustness, and provides an efficient and reliable model adaptive updating solution for edge video inference in smart cities, industrial detection, intelligent transportation and the like.

The present invention proposes a dual-arm robot wiring control method and system based on active perception, which relates to the field of robot control technology. It includes constructing a high-level primitive selection imitation learning network, identifying the primitive that needs to be executed currently based on the local image, the global image and the position of the end effector of the operating robot arm; inputting the local image and the position of the end effector of the operating robot arm into the low-level primitive imitation learning network model, obtaining the next action of the operating robot arm, and updating the position of the end effector of the operating robot arm; inputting the global image, the position of the end effector of the perceived robot arm and the cable point cloud data into the active perception network model, obtaining the next action of the perceived robot arm, and updating the position of the end effector of the perceived robot arm; and looping the above steps based on the updated information to complete the entire wiring process. The present invention adopts an active perception method to reduce the self-occlusion rate of the robot operating deformable objects and improve the robot's autonomous wiring capability.