37 results about "Point localization" patented technology

The invention relates to the technical field of the computer applications and the computer view, in particular to a multi-view two-dimension facial feature point automatic positioning method. The multi-view two-dimension facial feature point automatic positioning method comprises training and testing stages; the training stage comprises the following steps of firstly dividing a training data set containing a multi-view facial image into a plurality of training sub-sets; secondly training a single-view feature point positioning engine, namely training a cascading regression feature point positioning engine aiming at every training sub-set in every step. The multi-view two-dimension facial feature point automatic positioning method improves the robustness of feature point positioning under every view and can detect feature points of two-dimension facial images with the horizontal deflection angle from minus 90 degrees to 90 degrees.

As that invention is applicable to the field of computer technology, A facial feature point localization method based on cascade regression is provided, a device, an apparatus and a medium are provided. The method comprises: Through the multi-scale convolution neural network in the trained cascade regression model, predicting a feature point position set of a face image to be located and extracting a global feature of the image, according to the global feature, modifying the feature point position set of the face image to be located by each refinement layer in the cascade regression model, according to the corrected feature point position set, the final position of all feature points on the face image to be located is determined, Among them, the cascade regression model is based on the Jacobian matrix in-depth optimization strategy training, Each refinement layer includes a local feature extractor, a feature fusion device and a regressor, which improves the effect of image texture information utilization and model training in the process of facial feature point positioning, and further improves the accuracy and accuracy of facial feature point positioning.

The invention provides an asphalt pavement pressure generator embedding method and belongs to the technical field of highway engineering piezoelectric pavement paving. The method includes the steps that basic point localization, fixed point pit excavation, circuit slotting, pressure generator embedment, wire embedment and antiskid coating laying. A road section suitable for pressure generator embedment is selected, two fixed point piles A and B are driven to the positions outside a road shoulder of the road section, the distance dA between each point and the point A and the distance dB between each point and the point B are measured and recorded, after a face layer of the road section is laid, the point A and the point B are measured, compared with data recorded previously, reserved embedding points of pressure generators are found, a pit is excavated in each embedding point, and each pressure generator is placed in the corresponding pit, cushioned flatly and fixed through asphalt mastic and then encapsulated through a gap sealing and filling material. The gap sealing and filling material is obtained by compounding SBS modified asphalt and compound modified kieselguhr. Embedding links are tightly combined, a construction technology is simple, the pressure generators can be effectively fixed, damage to a pavement is greatly reduced, and embedding work of the pressure generators is effectively guided.

The invention provides a binocular vision-based hand key point space coordinate acquisition method, which comprises the following steps of: firstly, performing three-dimensional calibration on binocular cameras, and establishing coordinate system conversion models, namely acquiring internal and external parameters and distortion coefficients of the cameras and a rotation and translation matrix between the two cameras; secondly, preprocessing the video shot by the binocular camera, wherein the preprocessing comprises cutting, distortion correction and the like; processing the video frame by frame by using a machine learning pipeline, and obtaining pixel coordinates of 21 key points of the hand; and finally, calculating the real coordinates of the hand key points in the three-dimensional space by adopting a least square method based on an optical axis convergence model. According to the method, binocular vision information is utilized, the three-dimensional space coordinates of the 21 hand key points containing all the joint points are accurately positioned and recovered by simulating the human eye structure, hand form reconstruction is more accurate, and accurate technical support is provided for application research of hand key point positioning in human-computer interaction.

The invention discloses an interactive LED display device, which comprises a control terminal and at least one control system which is connected to the control terminal, wherein the at least one control system comprises a unit information acquisition sub-system, an LED display sub-system and a recognition and induction sub-system which are sequentially electrically connected. According to the interactive LED display device provided by the invention, when a sensor in the LED display device is touched and triggered by a human body, induction signals are returned to the control terminal via the sensor, and the induction signals are located via the control terminal and the unit information acquisition sub-system, so that corresponding playing data can be controlled to play via an LED display screen module, a real-time response effect can be achieved and response-point localization can be implemented; and an always-on-display state is kept or specific videos are played when the LED displayscreen module is not triggered. When the induction signals of the LED module are touched by the human body or an object, corresponding play contents or specific functions of a touch point are startedvia the control terminal. Therefore, the LED display device and display contents can be controlled by a person in a touch mode, and diversity of the functions and display contents of the LED display device can be enhanced.

The invention discloses a stereotaxic four-point pigeon brain localization device comprising a three-dimensional electrode moving device fixed on a baseplate, a left ear rod holder and a right ear rod holder. The left and right ear rod holders are identical in structure, and each of the left and right ear rod holders comprises a column, an ear rod and a side plate connected to the ear rod; a coracoid bar is connected between the side plates of the left and right ear rod holders. The invention further discloses an operation method of the stereotaxic four-point pigeon brain localization device. The stereotaxic four-point pigeon brain localization device utilizes the lever principle and adopts a four-point localization method, namely two ears are localized and fixed with the ear rods, the coracoid bar is moved to the position between the nose and the coracoid corner of a pigeon to be localized and fixed, and then a pigeon coracoid adapter is butted against the part between the nose and the upper coracoid tip to be localized. Due to the fact that force of the coracoid bar is outward and force of the adapter is inward, the upper jaw bone of the pigeon is firmly clamped between the coracoid bar and the adapter just like a lever, the pigeon brain is reliably localized and fixed, and repeated localization accuracy is high.

The invention discloses a training method of a human body key point positioning model, a human body key point positioning method and device, equipment, a medium and a program product, and relates to the field of artificial intelligence, in particular to deep learning and computer vision technologies. The implementation scheme comprises: determining an initial positioning point of each key point by using a first sub-network of a positioning model; obtaining N candidate points of each key point according to the position of the initial positioning point; extracting a second feature graph of the sample image by using a second sub-network of the positioning model, and forming N groups of graph structure feature data according to the second feature graph and the candidate points; performing graph convolution on each group of graph structure feature data by using a second sub-network to obtain the offset of each key point of N groups; correcting the initial positioning point by using the offset to obtain positioning results of the N groups of key points; and respectively calculating loss values according to the marked truth value of each key point and each group of positioning results, and performing supervision training on the positioning model according to each group of loss values. According to the invention, the accuracy and robustness of key point positioning are improved.

The invention provides a portrait caricature generating method, which comprises the steps of S1, performing face detection and feature point localization on an input image; S2, preprocessing the inputimage according to a face detection and feature point localization result so as to be aligned to a reference face image, and obtaining a feature point set of an input face aligned to the reference face image; S3, comparing with reference face features, and calculating personal features of the input face; S4, exaggerating the personal features under a deformation constraint criterion to obtain a portrait caricature feature point set of the input face; and S5, establishing triangular coordinate systems respectively based on the face feature point set of the input face after being aligned to thereference face image and the portrait caricature feature point set, realizing image deformation between the two corresponding triangular coordinate systems, and generating a portrait caricature. Theportrait caricature generating method is based on the triangular coordinate systems, does not need training samples, can generate a portrait caricature quickly by only requiring a reference face image, and can be applied to a quasi-real-time system.

The invention discloses a burn-through point positioning method applicable for an updraft sintering machine and a positioning system thereof. Atmospheric temperature is detected by temperature thermocouples which are arranged on a petticoat pipe close to a batch layer, the temperature thermocouple with the highest temperature and two adjacent temperature thermocouples are found out, and the burn-through point position is calculated by utilizing the spline interpolation function according to reference positions of the three atmospheric temperatures. By using the positioning method and system, the burn-through point positioning is more precise than sinter point positioning of down-draft sintering. At the same time, the positioning method and system have good functions on the lumpiness of lump sinters, rolling drum rate and desulfurization rate, effectively improve smelting quality of blast furnaces in subsequent procedures, save energy and lower consumption.

The invention discloses a machine vision-based driver fatigue detection method. The method includes the following steps of: S1, facial image acquisition; S2, eye region localization; S3, eye movementextraction; and S4, fatigue state detection. According to the method of the invention, an eye region guiding feature point localization algorithm is adopted to realize the localization of eye region feature points; in an eye region image where eye locations are determined, the physiological structural features of irises are adopted to locate the center locations of the irises; and therefore, the imaging difference of the eye image under different illumination conditions can be avoided, the adaptability of a detection system to illumination can be improved, and the influence of a driver's limbmovement and a nighttime sunglasses working condition on fatigue state detection can be avoided.

A method for determining whether a point is contained in a sub-volume of a digitized medical image, includes providing a tetrahedron volume mesh (TVM) representing a sub-volume of a digital image volume and a point M, finding a vertex P of said TVM that is closest to point M, finding a tetrahedron T_i in said TVM that defines a solid angle around point P where point M is located, wherein if M is inside T_i, then M is inside the TVM, if M is not inside T_i, finding a facet F of tetrahedron T_i through which a line PM connecting points M and P exits T_i, and determining whether a next tetrahedron T_i+1 of said TVM exists along line PM, wherein if no such tetrahedron T_i+1 exists, then point M is outside the TVM, and wherein if T_i+1 exists, determining whether point M is inside tetrahedron T_i+1.

The invention discloses a machine vision-based multi-scene iris region localization method. The method comprises the following steps of: S1, facial image acquisition, S2, facial region determination;S3, eye region localization; and S4, iris region localization. According to the method of the present invention, an eye region guiding feature point localization algorithm is adopted to realize the localization of eye region feature points; in an eye region image where eye locations are determined, the physiological structural features of irises are adopted to locate the center locations of the irises; and therefore, the imaging difference of the eye image under different illumination conditions can be avoided.

The invention provides a spatial multi-point positioning method, device and equipment based on multi-line fitting and a medium, and the method comprises the steps: controlling a manipulator to move along a Z axis, and determining a first edge point/a second edge point according to the distance between the manipulator and a first positioning bump/a second positioning bump detected by a laser sensor; controlling the manipulator to move a first preset step length along the X axis until a first edge point group/a second edge point group is obtained, and controlling the manipulator to return to the initial measurement point; the manipulator is controlled to move along the X axis, and a third edge point/a fourth edge point is determined according to the distance, detected by the laser sensor, between the manipulator and the first positioning convex block/the second positioning convex block; controlling the manipulator to move a second preset step length along the Z axis until a third edge point group/a fourth edge point group is obtained; and according to the first edge point group, the second edge point group, the third edge point group and the fourth edge point group, determining position coordinates of the thrombus instrument placement platform. According to the scheme, accurate positioning of the movement point position in the three-dimensional space is achieved.

The invention discloses an instrument measuring point installation intelligent design method which comprises the following steps: acquiring a pipeline wiring design drawing, and acquiring category attributes of measuring point attributes of pipelines at different positions according to the pipeline wiring design drawing; establishing a pipeline measuring point instrument type database according to the category attributes of the pipelines; according to the measuring point attributes of the pipelines, establishing a measuring point installation pipe seat model and an instrument model database; establishing a measuring point positioning mathematical model according to a flowing medium in the pipelines and a preset installation specification; and inputting the measuring point instrument type database, the measuring point installation pipe seat model database and the measuring point positioning mathematical model into process pipeline three-dimensional model software, and automatically generating an instrument measuring point design file. According to the method, the problems that in an instrument measuring point installation design method, measuring point instrument setting information on a flow chart pipeline and pipeline three-dimensional design are mutually independent, and no specific measuring point positioning exists in a design file can be solved.

The invention discloses a method for positioning the position of a slow sql mapping code through a section technology, which comprises the following steps of: 1, identifying a section: in a scene of a Java application slow sql, identifying a section point by all types of methods under a Mapper or Relocation packet of a persistent layer; and 2, section disposal is realized, execution of the method is intercepted by selecting an Around section surrounding mode, then a data basis can be provided for next analysis work, time marking is performed before and after execution of the intercepted method, and consumed time for execution of jointPoint. Process () is calculated. According to the method, hidden danger points can be quickly identified, and the positioning level is a specific code line; according to the method, scene data (information such as code positions, incoming parameters, outgoing parameters and data sizes of the outgoing parameters) of problems can be provided, and problem analysis and positioning are facilitated; the implementation and monitoring early warning scheme of the invention can be applied in a test environment, so that hidden danger problems can be found in advance, and the influence on the stability of a production environment is reduced.