179 results about "Occlusion detection" patented technology

The invention relates to the technical field of marine monitoring, in particular to a marine target visual servo tracking method, which comprises the steps of multi-modal sensor fusion, a self-adaptive visual tracking algorithm, a servo control and visual collaboration mechanism and a shielding processing and target re-identification strategy. The problem that tracking is unstable under the conditions of illumination change, ship body shaking, target shielding and the like in a traditional method is solved. The IMU, the GNSS and the visual data are fused through extended Kalman filtering, ship body shaking is compensated, and the target state estimation precision is improved; the improved D-Fi ne target detection model is combined with an online feature updating mechanism to dynamically adapt to the appearance change of the target; the prediction and correction control strategy and the double-closed-loop PI D controller cooperate to adjust the camera holder, and the tracking delay is reduced; the multi-clue shielding detection and space-time joint feature matching technology ensures accurate re-identification of the target after shielding is removed. The real-time performance and robustness of the tracking system on an embedded platform are improved, and an efficient and stable target tracking solution is provided.

A dynamic adaptive depth camera occlusion detection method, system and device, and a storage medium, the method comprising: step T1: constructing a three-frame time sequence analysis window, continuously obtaining infrared speckle image sequences at different moments and corresponding dense depth map sequences, and establishing a pixel-level space-time alignment mapping relationship; t2, calculating a motion vector between adjacent frames, and executing the step T3 when the average motion intensity exceeds a dynamic threshold value; wherein the dynamic threshold is determined by the dense depth map; t3, positioning a speckle feature point in the infrared speckle image, generating an analysis window by taking the speckle feature point as a center, calculating a gradient change rate, and when the gradient change rate is smaller than a gradient threshold value, marking the region as a primary candidate region; and T4, carrying out space-time consistency comparison on the primary candidate region and two adjacent frames, and reserving a shielding region of which the variation value is smaller than a variation threshold value in three continuous frames. The method is high in precision, good in real-time performance and high in stability.

The invention belongs to the technical field of augmented reality (XR) and computer vision crossing, relates to fusion application in immersive digital performance, and provides a virtual actor reconstruction and immersive presentation system based on 4D Gaussian splash modeling and XR space positioning. The system comprises a set of spherical multi-camera-position high-synchronization camera shooting matrix used for capturing dynamic images of actors; performing dynamic modeling on the multi-angle image through a 4D Gaussian splashing technology, and outputting a virtual actor point cloud model which can be used by XR equipment; vPS visual positioning and an SLAM tracking module are combined, and precise mapping positioning of a performance space is achieved in AR / MR equipment. The method supports the immersive watching of the actor image at the audience end at a 360-degree free visual angle, and realizes the natural presentation of the virtual actor without dead angles and wearing in cooperation with shielding judgment and a real-time rendering engine. The system is widely applicable to on-site entertainment scenes such as immersive theaters, text travel performances, brand activities, concerts and television programs.

Disclosed is a image processing apparatus and a method for controlling the image processing apparatus. The image processing apparatus according to an embodiment of the present disclosure may identify an object from an acquired image, determine whether the object is hidden by another object by using an aspect ratio of a bounding box of the detected object, and based on the object being hidden, estimate an entire length of the object based on coordinate information of the bounding box. Accordingly, the size information of the hidden object may be efficiently identified while a large amount of database is applied or resources of the apparatus is minimized. The present disclosure may be in connection with a surveillance camera, an automotive driving vehicle, an artificial intelligence module of at least one of a user terminal or a server, a robot, an augmented reality (AR) device, a virtual reality (VR) device, a device related to a 5G service, and the like.

The invention relates to the technical field of mine safety monitoring, in particular to an anti-shielding real-time deformation monitoring method and system for roof fall and wall caving disasters. The method comprises the following steps: extracting a bounding box and a mask of a discontinuous region from roof caving and rib spalling point cloud data, generating complete point cloud data, and analyzing a preliminary deformation index; extracting a track list of the shielding target, a prediction shielding area, a deformation trend and a deformation prediction value; in combination with the bounding box, the mask and the initial deformation index, obtaining a displacement vector and curvature change of the repaired missing region; and in combination with the trajectory list, the predicted shielding region and the deformation characteristics after compensation completion of the initial deformation index, obtaining a compensated and corrected displacement vector and curvature change, and further evaluating a deformation state and an early warning level. The method combines shielding detection, dynamic object tracking, multi-view fusion and deep learning completion to realize anti-shielding deformation feature extraction, time sequence trend prediction and accurate early warning, and provides reliable technical support for mine safety production.

The invention relates to a multi-source perception-based mechanical arm operation data acquisition system and method, the system comprises an isomorphic mechanical arm operation module, a multi-view visual image acquisition module, an optical motion capture module and a shielding detection and data fusion module, vision, pose and force sense data are cooperatively acquired through multiple sensors, and time-space consistency is synchronously kept by adopting hardware. The system has a multi-dimensional occlusion detection mechanism and can dynamically switch data processing strategies according to the occlusion degree: visual data is preferentially adopted when no occlusion exists, multi-sensor information is fused when partial occlusion exists, and force feedback compensation is predicted and introduced based on a dynamic model when complete occlusion exists. Physical feedback and visual perception are creatively combined, the problem of data missing of a traditional single sensor data acquisition system under the shielding condition is solved, more comprehensive and more reliable data support is provided for mechanical arm algorithm training and performance optimization, and the method is particularly suitable for complex operation scenes with shielding.

The invention discloses an unmanned aerial vehicle inspection optimal waypoint discovery method based on three-dimensional space shielding detection, and the method comprises the steps: 1) obtaining the latitude and longitude coordinates and height information of a to-be-inspected point, and loading a triangular patch model of a surrounding building or terrain; 2) generating a plurality of candidate observation waypoints with different heights and orientations above the to-be-inspected point; 3) calculating whether the sight line from the candidate waypoint to the ground inspection point is shielded by a triangular patch or not by using an R tree spatial index and a ray tracing algorithm; and 4) comprehensively considering the sight shielding condition, the observation distance and the observation angle, and selecting a waypoint which meets the non-shielding condition and has the optimal observation effect as an unmanned aerial vehicle inspection position. Compared with the prior art, the method has the advantages of being high in efficiency, low in cost and good in quality, the problem of viewpoint shielding in unmanned aerial vehicle routing inspection path planning under the urban complex environment is effectively solved, and the routing inspection efficiency and the data acquisition quality are improved.

The invention provides a lens shielding detection method, image acquisition equipment and a storage medium. The method comprises the following steps: segmenting a target image to obtain at least two block images; each block image has the same size and shape; obtaining the regional high-frequency confidence degree of each block image corresponding to the predetermined three components; obtaining a region bright color ratio corresponding to each block image; the regional bright color ratio is the ratio of pixel points with the brightness greater than the preset brightness in the block image; obtaining a region motion probability corresponding to each block image; the region motion probability is color brightness change in the block image; according to the current region high-frequency confidence coefficient, the historical region high-frequency confidence coefficient, the region bright color duty ratio and the region motion probability, the region shielding probability is obtained through calculation; and determining a lens shielding state according to the region shielding probabilities corresponding to the plurality of block images. The accuracy of detecting whether the lens is shielded or not can be improved.

The present application provides a method and device for window occlusion detection, which belongs to the field of sensor technology and can be used for assisted driving and autonomous driving. The method includes: obtaining a first window echo intensity and a second window echo intensity, the first window echo intensity represents the reference window echo intensity, and the second window echo intensity represents the window echo intensity at the current moment; obtaining the detection parameters of the target point at the first moment and the detection parameters at the second moment, the detection parameters including the reflectivity and / or position coordinates of the target point; according to the first window echo intensity, the second window echo intensity, the detection parameters at the first moment and the detection parameters at the second moment, obtaining a window occlusion parameter, the window occlusion parameter is used to indicate the window occlusion state. This method improves the accuracy of lidar window occlusion detection, so as to improve the terminal's advanced driving assistance system (ADAS) capabilities in autonomous driving or assisted driving.

A technical solution for monitoring operation of a medical delivery device such as a wearable patch pump for occlusion is provided that employs sensing pump motor current during at least one of a dispense stroke and an aspirate in a pump cycle. The solution can be implemented alone or in combination with other occlusion sensing methods that use one or more of pump measurement data such as pump stroke duration (e.g., duration of aspirate stroke or dispense stroke in a rotational metering-type pump or a reciprocating-type pump), end-stop or limit switch activation, and duration difference between aspirate and dispense strokes to detect occlusion. Accuracy of occlusion detection based on measured data in a fluid delivery device is improved by using dead band normalization of loaded measured data to unloaded measured data.

The invention relates to the technical field of paris polyphylla plant disease and insect pest data processing, and discloses a paris polyphylla planting plant disease and insect pest control scheme data analysis system and method. According to paris polyphylla plant growth image shelter type identification information, an intelligent search algorithm and standard preset paris polyphylla plant growth image shelter removal algorithm type information are combined to carry out shelter feature removal algorithm type accurate analysis in a paris polyphylla plant growth image; scientific matching of image shelter removal algorithm objects based on shelter interference feature types in paris polyphylla disease and insect pest monitoring images is realized, and the quality of paris polyphylla disease and insect pest monitoring image acquisition is improved; the method can autonomously and efficiently remove interference features of plant surface attached shelters in the paris polyphylla plant growth image, realizes efficient and accurate processing of the interference features of the shelters in the paris polyphylla plant growth image, improves the authenticity of paris polyphylla plant disease and insect pest monitoring image acquisition, and improves the paris polyphylla planting disease and insect pest control effect.

The present invention discloses a panoramic three-dimensional instance recovery method, system, terminal and storage medium based on a single image. The method includes: extracting two-dimensional information of a single RGB image, feature up-dimension based on the two-dimensional information, refinement and instance grouping of three-dimensional semantic information, multi-level occlusion detection and occlusion removal, and foreground and background synthesis of the panoramic three-dimensional instance, so as to obtain a complete three-dimensional scene reconstruction result. The present invention realizes the panoramic three-dimensional reconstruction of indoor scenes in an occluded environment by using multi-level foreground and background segmentation and occlusion removal technologies, can automatically remove occluded areas, and obtain an image after removing occlusions.

Some embodiments relate to driver or occupant monitoring using vision language models (VLMs). Any number of DNNs in a detection pipeline may be replaced with a VLM, and the VLM may be prompted to determine whether a corresponding feature is present in an image or sampled frames from a video. To facilitate using the VLM(s) to control one or more downstream actions, the VLM(s) may be prompted using structured inputs, and a designated output format for a corresponding structured output may be enforced in any suitable manner. As such, any number of VLMs may be used to perform any number of driver and / or occupant monitoring tasks (e.g., driver drowsiness detection, driver distraction detection, driver or occupant out-of-position detection, driver or occupant identification, seatbelt usage detection, occupant presence detection, occupant classification, child presence detection, gesture recognition, occlusion detection, and / or others).

The invention discloses a three-dimensional human body posture estimation method and device based on multi-mode and occlusion perception, and relates to the field of computer vision, and the method comprises the steps: S1, obtaining a multi-mode video sequence comprising an RGB sequence, a depth sequence and an infrared sequence; s2, carrying out preprocessing to obtain a multi-modal video sequence with aligned data; s3, constructing a three-dimensional human body posture estimation model comprising a spatial-temporal feature extraction network, a shielding detection network, a time sequence coordinate classification module, a Z-axis depth prediction module and a three-dimensional coordinate splicing module; s4, training a three-dimensional human body posture estimation model by using the data-aligned multi-modal video sequence; and S5, performing three-dimensional human body posture estimation by using the trained model. According to the method, through RGB, depth and infrared three-mode fusion, X-axis and Y-axis coordinates are predicted by adopting a time sequence coordinate classification method of occlusion perception, and Z-axis depth prediction based on bucket separation and multi-mode and multi-resolution probability fusion is adopted, so that the problem of three-dimensional human body posture estimation in a complex occlusion scene can be effectively solved.

An embodiment of the present application relates to an occlusion detection method and a terminal device for a TOF camera. The method includes: obtaining an image generated by an image sensor according to the received first reflected light; preliminarily determining whether the TOF camera is occluded according to a first frame image generated by the image sensor when the projection light source is turned on and a second frame image generated by the image sensor when the projection light source is turned off; if it is preliminarily determined that the TOF camera is occluded, obtaining a voltage generated by a photoelectric conversion component according to the received second reflected light; if the voltage is greater than a preset voltage, determining that the TOF camera is occluded. In the embodiment of the present application, the terminal device preliminarily determines whether the TOF camera is occluded according to the first frame image and the second frame image, and then further determines whether the TOF camera is occluded by comparing the voltage generated by the photoelectric conversion component with the preset voltage. This method can accurately detect whether the TOF camera is occluded.

The application relates to the field of oral medicine diagnosis and discloses an AI-enhanced real-time occlusion detection system based on intraoral images, which comprises a multi-source intraoral image acquisition module, which is used for acquiring dynamic oral image data through a high-definition oral endoscope, recording the three-dimensional structure of a tooth surface in combination with a laser scanning device, and collecting the force distribution of an occlusion contact point through a pressure sensor array; the multi-source intraoral image acquisition module comprises a time stamp synchronization unit, which performs multi-source synchronization on the oral image frame sequence and the occlusion force distribution data by using a unified time mark; through the multi-source intraoral image acquisition module, the synchronous acquisition of dynamic images, three-dimensional structure data and occlusion force distribution is realized, and the time sequence consistency of the data is ensured through the time stamp mark. Compared with the acquisition mode of a traditional single data source, multi-source acquisition can comprehensively record the dynamic characteristics in the occlusion process, provide rich and accurate input for subsequent AI analysis, and avoid data omission or acquisition deviation.

An occlusion detection system to perform operations that include: capturing image data that depicts an environment at a client device, the environment including a target object at a position within the environment; causing display of a presentation of the environment at the client device, the presentation of the environment including a display of the target object at the position within the environment; detecting a first attribute of the display of the target object at the client device; performing a comparison of the first attribute of the display of the target object and a second attribute associated with the target object; and detecting an occlusion based on the comparison.

The invention discloses an infrared speckle depth camera occlusion detection method, system and device, and a storage medium. The method comprises the following steps: S1, obtaining a first infrared speckle image and a first dense depth map at a first moment; s2, extracting a first speckle center on the first infrared speckle image, calculating a first difference value between the first speckle center and surrounding pixels, and marking a first alternative area; s3, acquiring a second infrared speckle image and a second dense depth map at a second moment; s4, subtracting the first dense depth map from the second dense depth map to obtain a third dense depth map, and if the infrared speckle depth camera is in a motion state, executing the step S5; s5, extracting a second speckle center on the second infrared speckle image, calculating a second difference value between the second speckle center and surrounding pixels, and marking a second alternative area; and S6, marking the shielding area.

The invention discloses a method for training a detection model for cerebrovascular occlusion detection and a related product. The method comprises the following steps: performing deboning operation based on CT perfusion imaging data to obtain target image data containing brain tissues and blood vessels; performing vascular occlusion region labeling on the target image data to obtain a three-dimensional occlusion region label; extracting a blood flow arrival time graph based on the CT perfusion imaging data; based on the target image data and the blood flow arrival time graph, extracting a blood vessel region feature volume after weighted fusion under blood flow arrival time gating; inputting the blood vessel region characteristic volume into the detection model to carry out cerebrovascular occlusion detection; and calculating a loss function based on the three-dimensional occlusion region label so as to train the detection model for cerebrovascular occlusion detection. By using the scheme of the invention, the accuracy, robustness and clinical applicability of occlusion detection can be improved.

The invention relates to the technical field of intelligent livestock management, in particular to an intelligent monitoring method and system for rumination behaviors of cattle individuals, and solves the technical problem of low monitoring accuracy in the prior art. The method comprises the following steps: acquiring cattle herd monitoring video data, and performing cattle mouth part identification on image frames in the cattle herd monitoring video data through a target identification model to obtain a cattle mouth area image in each image frame; performing shielding detection on the cattle mouth area image in each image frame, and screening out a target cattle mouth area image from the cattle mouth area images in each image frame according to an obtained shielding proportion; processing the screened target cattle mouth area image through an optical flow algorithm, and generating an optical flow field sequence representing cattle mouth movement; and inputting the optical flow field sequence into a preset classifier, judging whether the cattle is in a rumination behavior, and outputting a monitoring result of the rumination behavior.

The invention discloses a human body posture estimation method with occlusion detection correction and kinematics constraint optimization functions, which comprises the following steps: acquiring motion image data, detecting and marking unreliable key points of a target skeleton in the moving image data in each frame of static image by confidence analysis, time sequence information, space consistency and biomechanical prior information, so as to reserve reliable key points of the target skeleton in each frame of static image; correcting the reliable key point position of the target skeleton in each frame of static image based on polynomial fitting, a symmetry inference strategy and a human skeleton propagation strategy to obtain an initial overall skeleton; and introducing multiple physical constraints to construct a kinematics constraint function, and performing iterative optimization on the initial overall skeleton by taking minimization of observation deviation as a target to output human body posture estimation. The method provided by the invention can effectively solve the problems of key point loss, unreasonable structure and dynamic instability in complex scenes such as shielding and rapid movement.

The utility model relates to a detection and measurement device facilitating shielding detection. The detection and measurement device comprises a detector supporting frame, a sliding rail, a calibrated scale and a shielding assembly. The number of the detector supporting frames is two, the two detector supporting frames are oppositely arranged, a gamma detector and a beta detector are arranged on the two detector supporting frames respectively, the emission direction of the gamma detector is right opposite to the emission direction of the beta detector, and the detector supporting frames are arranged on the sliding rail in a sliding mode. The shielding assembly comprises a shielding plate and a shielding installation part, the shielding installation part is of a U-shaped structure, installation grooves are formed in the two ends of the inner side of the shielding installation part respectively, the shielding plate is matched with the installation grooves, the shielding plate is embedded in the installation grooves, and the gamma detector and the beta detector are spaced. The shielding plate can be conveniently installed on the installation groove of the shielding supporting frame, so that the shielding plate is located between the beta detector and the gamma detector, and gamma rays emitted from a radioactive source are prevented from directly entering the beta detector.

The invention relates to the technical field of computer vision, and provides a shielding detection method, electronic equipment and a computer readable storage medium, and the method comprises the steps: obtaining a to-be-detected video stream which comprises multiple frames of to-be-detected images; and based on a preset video classification model, determining the occlusion probability of the plurality of frames of to-be-detected images, the occlusion probability being used for representing whether the plurality of frames of to-be-detected images have picture mutation. And if the occlusion probability is greater than or equal to a preset probability threshold, target detection results corresponding to multiple frames of first acquisition images acquired in a first time period are acquired, and the multiple frames of first acquisition images are multiple frames of continuous images acquired by the image acquisition device at a preset image sampling rate. And according to the target detection result, determining a shielding detection result corresponding to the multiple frames of to-be-detected images. According to the method and the device, occlusion detection can be performed on the to-be-detected video stream in real time, the false alarm rate and the missing report rate of the detection result are reduced, the accuracy of the occlusion detection result is improved, and the safety of a monitoring system is improved.

The application provides a camera shielding detection method and device, electronic equipment and readable storage medium, and relates to the field of image recognition. The method comprises the following steps: acquiring a target image collected by a target camera, and calculating the light intensity of the target image; based on the comparison result of the light intensity of the target image and a preset light intensity, using a target model to analyze the target image, and generating an analysis result; wherein, the preset light intensity is the average value of the light intensity of the image whose camera shielding judgment is wrong; and the analysis result is used to indicate whether the target camera is shielded. The camera shielding detection method, device, electronic equipment and readable storage medium provided by the application can timely judge whether the camera is shielded under any light condition according to the image frame collected by the camera in real time, and improve the accuracy of shielding detection.

Systems and techniques are described herein for occlusion detection. For instance, a method for occlusion detection is provided. The method may include generating a plurality of voxels in a voxel space based on a point-cloud representation of a scene; generating an annotation point in the voxel space based on sensor data representative of an object in the scene; projecting a ray from a sensor position in the voxel space to the annotation point; and determining whether the annotation point is occluded based on the ray and the plurality of voxels.

The invention discloses a food image recognition and calorie detection method based on deep learning, and belongs to the field of image recognition algorithms. According to an existing food and food recognition method, when food is shielded, feature loss is caused, the problems of noise, local aliasing and the like are included, and therefore extraction and matching of a recognition algorithm on facial features are interfered. A food image recognition and calorie detection method based on deep learning comprises the following steps: establishing a data set of simulated food images, and performing preprocessing; the preprocessing comprises image shielding processing, food target detection data set preprocessing, and creation of a data set required by a shielding detection network; determining a detection network and an image filling network for detecting the food image with the shielding condition; constructing a judgment network used for judging whether the to-be-detected food image has a shielding condition or not; and constructing a detection network for detecting the food image with the shielding condition to obtain a final result. According to the invention, the identification precision is improved.

A method includes controlling a robotic system using a virtual boundary and based on optical tracking of a tracker, detecting a condition in which an optical detection device is occluded from detecting the tracker, and maintaining the virtual boundary if the condition persists for shorter than a predetermined period of time.

This invention discloses a method and device for detecting the attitude of a spherical motor rotor. The method includes: developing a feature plate with three different visual feature points (first, second, and third) and fixing it to the top of the rotor output shaft; developing a visual detection system for the spherical rotor attitude, wherein when the rotor output shaft is located on the Z-axis of the northeast-northeast coordinate system, an image acquisition module is installed directly above the second feature point and calibrated; collecting and labeling typical and atypical spatial motion image datasets of the spherical rotor with three degrees of freedom; designing and training a network model suitable for detecting the real-time attitude of the spherical rotor using the YOLOv5s framework; the image acquisition module acquiring a frame of image in real time, using the network model to detect the coordinates of the three different visual feature points in the image, and calculating the attitude of the spherical rotor based on the detected coordinates and the geometric relationship of the rotor's spherical motion. This invention solves the problems of needing to reset and failure to detect feature points due to long-term occlusion in visual spherical rotor detection.

The application discloses the technical field of power grid operation, and particularly relates to a power grid operation site face recognition method based on deep learning, which comprises the following specific steps: S1: collecting face image pairs of power grid operation personnel and performing pretreatment to obtain a power grid operation personnel multispectral image pair dataset D1; S2: performing processing on any second face image pair P2 = <I''1, I'2> E D1 to obtain a power grid operation personnel multispectral image pair dataset D2; S3: constructing a face recognition model dynamic adversarial multi-modal network DAM-Net of the power grid site under the occlusion condition of the operation personnel, which is used for identity recognition of the operation personnel of the power grid operation site under the face occlusion condition; and S4: training and updating parameters of each layer of the DAM-Net network constructed in S3 to obtain a trained DAM-Net network. The DAM-Net of the application introduces a deformable convolution fusion submodule and a space-time attention mechanism submodule through a DODM, and has the advantages of being capable of significantly improving the accuracy and robustness of occlusion detection.

The invention relates to the field of image processing, and provides a seedling three-dimensional reconstruction method and system under a dynamic shielding condition, which can avoid the influence of camera equipment and ensure the image synchronization across cameras by carrying out distortion correction and time synchronization calibration on multi-view seedling sequence images, and then carry out dynamic shielding detection to realize the three-dimensional reconstruction of seedlings under the dynamic shielding condition. According to the method, the occlusion area of the seedling leaf is accurately identified, the occlusion area is recovered according to image weighted fusion, the robustness of occlusion area reconstruction is improved, the spatial area of the leaf is accurately marked through time sensing point cloud fusion reconstruction, and the unfolding area of the leaf is enabled to accord with an actual scene. According to the method, the quality and the accuracy of three-dimensional reconstruction of the seedlings are improved.