68 results about "Head shoulders" patented technology

The invention discloses a people counting method and device based on head shoulder detection. The people counting method based on head shoulder detection comprises steps of S1 arranging a heat area and a base line in a monitor video, S2,performing head-shoulder detection on the heat area in the image frame of the monitor video, S3,determining whether the head and the shoulder exist according to the head shoulder detection result, if yes, executing the step S4, S4 setting a head-shoulder frame and maintaining or updating a tracking list and continuing to execute the step 6, S6, performing tracking on each head-shoulder frame in the tracking list to obtain a group of tracking frames, S7, determining whether the tracked head-shoulder frame in the current frame satisfies a people counting trigger condition, S8, starting up a counter and performing people counting according to the tracked head-shoulder frame. The invention discloses a people counting device based on the head-shoulder detection. The people counting method based on the head-shoulder detection is wide in application range, high in detection, low in false detection rate and high in tracking performance.

The invention discloses a sitting posture detecting method and device, aiming at realizing sitting posture detection on a target user and reminding the target user of maintaining a standard sitting posture. The sitting posture detecting method provided by the invention comprises the following steps: acquiring an image of the target user, and detecting a face tilt angle, area of an eye region, and a head shoulder curve according to the image of the target user; comparing the detected result with preset standard values to obtain a comparison result; and sending a prompt message to the target user when the sitting posture of the target user is determined to be not consistent with the standard according to the comparison result.

The invention provides a video-based safety cap wearing condition detection method comprising the following steps: obtaining video images, and calculating two-frame gray scale difference values so asto find out a motion area; using an agglomerate detection method to find out a candidate area; carrying out deformable assembly model system detection for the candidate area, and finding a clear target head shoulder area; carrying out cascade system detection for the head shoulder area so as to find out the target head area; carrying out HSV color space conversion for the video image head area; using a nerve network to classify and determine the head area image under the HSV color space, and determining whether the target has worn a safety cap or not; using the HSV color space to determine thesafety cap color; finally outputting the head area and safety cap wearing detection result. The video-based safety cap wearing condition detection method is accurate in detection result, can reduce the detection difficulty, and can improve the detection efficiency.

The invention discloses a head-shoulder sequence image segmentation method based on double-pattern matching and edge thinning, and the method is carried out as per the following steps: step 1, determining a face position, namely, adopting the Bayesian risk decision to determine the face area by considering the distribution of the face on a color plane; step 2, determining the head-shoulder area, namely, considering the head-shoulder area as a combination of two rectangular areas, one as head rectangle and the other as shoulder rectangle, setting the width of the shoulder rectangle as three times of the width of the head rectangle, and finally determining the head rectangle through adopting two rectangular moving templates and taking the proportion falling in the two template areas as the matching standard; and step 3, edge thinning, namely, adopting the Canny edge detection operator to acquire the accurate contour of a moving object. The head-shoulder sequence image segmentation method has the benefits that the Bayesian risk decision mechanism is adopted to determine the face position, the double-pattern matching algorithm is adopted to further determine the head-shoulder area, and finally the edge thinning is conducted, as a result, the algorithm can efficiently segment the head-shoulder sequence.

An image analysis and deep learning-based number of people statistical method disclosed by the present invention comprises the following steps of A carrying out the pyramid model calculation on an input image, and generating the images of a plurality of resolutions and sizes; B sliding the windows on each layer of a pyramid, calculating the HOG feature values of the window areas, classifying via a linear support vector machine (SVM) classifier, and determining whether the windows are the head-shoulder areas; C for each head-shoulder area given out in the step B, extracting the corresponding image, normalizing to a set same size, and inputting in a deep neural network to obtain the classification output; D carrying out the non-maximum suppression on the all head-shoulder windows outputted in the step C to merge the overlapped detection results of the adjacent areas and scales. The image analysis and deep learning-based number of people statistical method of the present invention can improve the insufficiency of the prior art, and can realize a higher number of people statistical performance with the faster speed.

The invention discloses a method for counting pedestrians in a foresight monitoring scene. The method comprises: setting the positions of a detection region and a detection line, performing head region and head-shoulder region detection on a motion mask image in the detection region, and determining a definite head region according to a position relationship between the head region and the head-shoulder region; and tracking a motion track of the determined head region, and according to the motion track and the position of the detection line, counting the pedestrians by direction. The invention furthermore discloses an apparatus for counting the pedestrians in the foresight monitoring scene.

The invention discloses a method and device for recognizing the human shape in a smart home scene. The method comprises the steps of firstly capturing images in real time through a terminal camera, selecting a background image which does not contain a figure in the smart home scene, and performing preprocessing on the current frame and the background image; obtaining a human shape contour image according to a background subtraction method, filling holes in the human shape contour by using a morphological method, performing pixel distribution histogram and pixel gradient distribution histogramstatistics on the human shape contour image in the horizontal and vertical directions to obtain coordinate information of human shape feature points, building a head-shoulder model, calculating a head-shoulder ratio so as to be used for judging whether there is a human shape or not, calculating the shape feature based curvature, comparing the calculated curvature with a human shape curvature library for human shape recognition. If there is a stranger at present according to human shape recognition, the system sends alarm information to a client and transmits the current image to the client. The method and system disclosed by the invention can be widely applied to the smart home scene, bring comfortable and safe enjoyment to users and have very wide application scenes.

The invention proposes a pedestrian detection and tracking method based on a head-shoulder contour and a BP neural network. The method comprises the steps: firstly extracting a moving human body target in a video sequence through employing an adaptive mixed Gaussian background updating algorithm, and improving the background estimation precision through changing a learning factor of a mixed Gaussian model; secondly extracting an initial contour of an original target through employing a Canny operator, and carrying out contour clustering through combining a Mean shift algorithm, so as to obtain a completer body contour; thirdly building a head-shoulder contour model through combining a head-shoulder width-height ratio of a human body, extracting a head-shoulder contour characteristic vector, inputting the characteristic vector of the head-shoulder contour model into the BP neural network, clustering a plurality of human body head-shoulder models, and carrying out human body recognition; and finally tracking a detected pedestrian target through employing a particle filter. The method avoids misjudgment and wrong judgment because of the incompletion of a recognition target, improves the recognition accuracy of the pedestrian target, and reduces the calculation amount.

The invention provides a backlight photographing method and device, and the method comprises the steps: detecting whether a photographed scene is photographed in a backlight manner or not; determining the size of a head-shoulder region corresponding to a portrait of a photograph object in the photographed scene through head-shoulder detection if the photographed scene is photographed in the backlight manner; determining the distance from the portrait to a camera according to the size of the head-shoulder region, automatically focusing the head-shoulder region in the portrait according to the distance, and photographing and obtaining a first image; The method and device can effectively detect the head-shoulder region in the portrait in the backlight scene through head-shoulder detection, can achieve the automatic photographing of a clear image of the head-shoulder region in the backlight scene, and cannot cause a black face phenomenon because of poor exposure of a person.

The invention provides a driver phone call making / answering state identification method and device. The method comprises steps that a window of a target vehicle in a monitoring image is positioned; acandidate driver detection area is acquired according to the window positioning information and the license plate information of the target vehicle; the candidate driver detection area is detected through employing a direction gradient histogram and a support vector machine to acquire a head shoulder area of a driver; the head shoulder area is sequentially inputted to a first CNN network and a second CNN network, preliminary screening of the head shoulder area is carried out through the first CNN network to acquire a suspected phone call making / answering head shoulder area of the driver, the output result of the first CNN network is further screened by the second CNN network to acquire a phone call making / answering state of the driver. The method is advantaged in that false detection of non-real phone call making / answering under complex occasions can be eliminated, identification precision is improved, and scene adaptability and robustness are good.

The invention discloses a space personnel positioning device and a method based on a multi-depth camera, which utilizes a plurality of 3D intelligent depth cameras to work cooperatively to obtain 3D depth images of scenes in the space, and extracts human body objects after distortion correction and 3D depth image enhancement. Then, the depth histogram feature of the head-shoulder region and the spatial density feature of the head region are extracted according to the obtained human body target region. Combined with 2D image and 3D depth image of human head, the convex square similarity featureof head is used to judge whether the locating object is human or not. Then, the coordinates of the center point of the head and the height of the person are calculated by three-dimensional re-projection transformation. Finally, the positioning data of a plurality of depth cameras are weighted and fused and the final positioning information of the space personnel is determined. The invention has the advantages of strong real-time property, easy expansion, accurate data and no need of auxiliary positioning facilities, and has wide application value in the field of area monitoring, VR positioning and human-computer interaction.

The invention provides a head shoulder image reconstruction method based on a model and an object in a low bit rate video call. The method is characterized in that with the combination of advantages of object-based coding and model-based coding, a low bit rate model-and-object-based hybrid coding method is provided; the method performs 2D deformable model automatic training of a face region and then coding; for a head shoulder region, a 2D grid formed based on a Delaunay algorithm is adopted, grid iteration is used to well match with a head image; then the image is subjected to JPEG2000 compression to obtain object texture, and an object grid and the object texture are sent to a reception end through a channel and then coding is shared. The method has the high coding efficiency via the grid segmentation and texture compression and achieves good user experience. According to the invention, transmission of head shoulder videos in the low bit rate is achieved, the experience effect after video image reconstruction is still good, and the utilization efficiency of communication resources improved.

The invention relates to a face-gesture cooperative verification method based on deep learning detection so that face and gesture detection can be completed by using an algorithm for a low-power-consumption device and thus an effect of face-gesture cooperative verification can be realized. Overall head-shoulder detection is carried out on an image reduced substantially by using a deep residual convolutional neural network; a face is detected based on a range with head-shoulder occurrence; and then a gesture is detected on the head and shoulder positions and then a trigger condition is verifiedbased on the detected face and gesture. Compared with the traditional gesture-based verification system, the face-gesture cooperative verification method has the following advantages: the stability is high; false triggering by a user and the out-of-order probability of the system are reduced substantially; the stability is improved; and the method can be applied to startup-shutdown verification of a mobile phone of a home appliance.

The invention discloses an intra-frame updating method for head-shoulder video sequences. The method comprises the following: 1) a step of performing face detection and verification on a face-updating frame image; 2) a step of coding the face-updating frame image, which is to code face-updating frames through a P-frame coding mode by taking a previous common P frame as a reference, perform mandatory intra-frame updating on a face region, namely to integrally update a face-region macro block in the current to-be-coded face-updating frame into an intra-frame coding mode, and update a non-face region according to a conventional intra-frame updating method; and 3) a step of coding a common-frame image, which is to adopt an inter-frame coding mode. The method has the advantages of preventing error diffusion of an error macro block in the face region and improving error control performance at a frame level and a macro-block level. Because the method specially protects the face region, in application occasions of video phones, program broadcast and other head-shoulder video sequence transmission, the method has better video-error recovering performance than the common updating method in an intra-frame coding mode.

The invention discloses a manufacturing process of an inner-hexagon cylindrical head shoulder screw. The manufacturing process comprises the following steps: cutting materials; intensively bunching; pre-upsetting; upsetting again and chamfering inside and outside; punching an inner hexagonal hole and molding; carrying out thread rolling; and carrying out heat treatment. By adopting the manufacturing process of an inner-hexagon cylindrical head shoulder screw, provided by the invention, the processing technology is simple and convenient; and a mould component is used for processing, so that the processing is convenient, the production efficiency is greatly improved, the concentricity, the parallelism and the luminance of each element of the processed screw are consistent, the size and precision requirements of the product are strictly ensured, and the screw strength is improved.

The invention discloses a pedestrian re-identification method based on color matching, comprising the following steps: a moving area in a video is detected to obtain a pedestrian candidate area; the preliminary pedestrian area is obtained by using head-shoulder detector for pedestrian candidate area, and the DPM model is used for pedestrian precision inspection, by tracking the detected pedestrians, the target tracking sequence is obtained, the pedestrian area is divided into torso and legs, the color descriptor of pedestrian is obtained by using the proposed color quantization method, and thetarget sequence information is stored in the database. The histogram intersection method is used to calculate the similarity between the query targets and the pedestrian sequences in the database, and the query results are returned in descending order of similarity. The method of the invention can stably detect and track pedestrians in different background environments and under different monitoring cameras, and the proposed color quantization method can effectively cope with the influence of different pedestrian postures, resolutions and illumination, and realize cross-camera searching and querying the target sequence similar to pedestrians.

A multifunctional bed capable of assisting a patient to roll over is disclosed. A roll-over board is symmetrically installed on both side of that middle line of the bed board, the roll-over board is divided into a head-shoulder part, a waist-hip part and a leg part, one end of the turning-over board is articulated with a rib plate at the middle line of the bed board, the other end of the turning-over board is connected with the edge of the bed board through a turning-over board turnover driving cylinder, and a turning-over buffer air bag is arranged in the gap between the turning-over board and the bed board. A pneumatic control box is arranged below the bed plate, an air pump is arranged in the box, an air pump outlet is connected with an air distribution valve, and all driving cylindersand buffer air bags are connected with the air distribution valve through a separate air duct. Four corners of the bed board are vertically uniformly distributed with four pillars. The left pillars and the right pillars of the middle line of the bed board are fixedly connected by longitudinal beams. Two cross beams are arranged between the two longitudinal beams. The cross beams can move along thelongitudinal beams. An upper / lower limb rehabilitation training module is hoisted on the cross beams. The module can move along the cross beams. After the invention is adopted, the nurse can help thepatient to turn over without personal assistance.

The invention relates to a sieve pipe self-locking switch controller. The sieve pipe self-locking switch controller is mainly composed of a conical sliding rail connector, a movable stirring block, a spring pressing cap, a reset spring and a front guiding head. Two to six dovetail grooves evenly distributed in the circumferential direction of the axis are machined in a conical structure body of the conical sliding rail connector. The movable stirring block is of a round flaw structure, and T-shaped sliding rails are machined on the bottom face, arranged in the dovetail grooves of the conical sliding rail connector and are in sliding fit. A rod head is machined at the front end of the movable stirring block, the top end of the rod head makes contact with one end of the spring pressing cap, and the spring pressing cap is arranged on a shaft outer sleeve of the front guiding head and is in sliding fit. The reset spring is arranged between the spring pressing cap and the front guiding head shoulder part. The end of the front guiding head is a conical body. The sieve pipe self-locking switch controller is used in cooperation with a sand control sieve pipe with a self-locking switch, one or more sand control sieve pipes at the assigned position under a well can be controlled according to the actual needs and can be made to put into production or quit production, and accordingly the purposes of flow adjusting, water controlling and production system optimizing are achieved.

The invention discloses a face detection method. The method comprises the following steps: constructing an image pyramid for an image to be detected; extracting aggregation channel characteristics ofeach layer of image of the image pyramid to obtain a characteristic pyramid of the to-be-detected image; obtaining a plurality of first detection frames of the to-be-detected image by using the firstsliding window, and classifying the first detection frames to obtain a plurality of candidate face frames; merging the candidate face frames; obtaining a plurality of second detection frames of the to-be-detected image by using a second sliding window, and classifying the second detection frames to obtain a plurality of candidate head-shoulder frames; combining the candidate header-shoulder framesand predicting a human face in the head-shoulder frames to obtain a predicted human face frame; and merging the merged candidate face frame and the predicted face frame to obtain a target face frame. The invention also provides a face detection device, a computer device and a readable storage medium. According to the invention, rapid high-detection-rate face detection can be realized.

The invention discloses a full-automatic double-head shoulder injection molding machine which comprises a rack, and a working platform I and a working platform II which are mounted on the rack, wherein a motor I and a circular turntable mechanism driven by the motor I are mounted on the working platform I; the full-automatic double-head shoulder injection molding machine is characterized in that aplurality of mandrel holders are arranged on the circular turntable mechanism; mandrels are arranged on the mandrel holders; a plurality of air blowing openings are formed in the mandrels; and an automatic pipe feeding device, an automatic injection molding device, an automatic forming device, a cooling device and an automatic pipe discharging device are arranged on the circular turntable mechanism in sequence. Compared with the prior art, the full-automatic double-head shoulder injection molding machine disclosed by the invention has the advantages that by full-automatic forming machining operation of a hose, the hose forming machining efficiency is improved, and the labor intensity is reduced; and by the adoption of the full-automatic pipe feeding device, two pipes are fed simultaneously, so that high-speed operation can be realized, the production efficiency is improved, the number of the used mandrels is reduced, the efficiency is improved, and the cost is reduced.

The invention relates to an anchoring device for a taper sleeve-locking reinforcements and an anchoring method thereof, and belongs to the technical field of reinforced concrete anchoring. The anchoring device comprises a plurality of clamping sheets and taper sleeve anchoring plates at the outer sides of the clamping sheets, wherein each clamping sheet is provided with a tooth-shaped inner surface and a single-conical-surface outer surface, and reinforcements to be anchored are wrapped and clamped through the tooth-shaped inner surfaces; each taper sleeve anchoring plate is in a head-shoulder integrated shape and provided with a conical hole at the centre, and the clamping sheets and the reinforcements to be anchored sequentially sleeve the inner centres of the conical holes and locked simultaneously through the taper sleeve anchoring plates; the tooth-shaped inner surface of each clamping sheet is provided with a plurality of spaced trapezoidal teeth, and threads for anchoring are not needed at the end parts of the reinforcements to be anchored. The anchoring device disclosed by the invention omits the prefabrication link of reinforcement thread machining, saves the machining structure and the machining cost of the prefabrication link, and is fast, convenient and reliable to install, and easy for control quality.

A vacuum booster detector at the push rod position has a pressure head which includes a pressure head convex to be inserted into the vacuum booster and connected with the the main body of the pressure head by a pressure head shoulder; there is a mandril within the pressure head and a first seal right is set at the contact surface between the mandril and the pressure head; a second seal ring is set at the pressure head shoulder; the pressure head flows on a linear sliding rail, on which there is no driving mechanism to drive the pressure head. in product detection, the convex of the pressure head enters into the front shell of the product firstly and the pressure head shoulder push against the outer wall of the front shell; under the driving of the driving mechanism, elastic deformation happens on the second seal ring and the pressure head is sealed with the front shell; then, the mandril moves and pushes the push rod and the reaction plate to reset and the push rod height can be measured by measuring the displacement of the mandril.

The invention provides a head and shoulder region positioning method and device. The method comprises the steps that a video image in a cab is acquired; the video image is input into a head and shoulder detection model; and a head and shoulder region in the video image is positioned through the output of the head and shoulder detection model, wherein the head and shoulder detection model formed by training a classifier through a number of head and shoulder training sets and a number of non-head-shoulder training sets. According to the invention, the problems of low project detection rate and low detection precision, which are caused by the influence of objective conditions such as ambient light and the like in the prior art, of head and shoulder region positioning are solved; the head and shoulder region is accurately and effectively positioned; and the robustness is great.

The invention provides a safety belt identification model establishing method and device. The safety belt identification model establishing method comprises the steps that a head-shoulder zone image is acquired, fastened safety belt training sets and un-fastened safety belts training set are obtained according to the head-shoulder zone, positive sample characteristics are determined according to the fastened safety belt training sets, negative sample characteristics are determined according to the un-fastened safety belt training sets, and a safety belt identification model is obtained by utilizing positive sample characteristic and negative sample characteristic training classifiers. The problem that the un-fastened safety belt behavior detection accuracy rate is low in the prior art is solved, establishment of the safety belt identification model is achieved, and the un-fastened safety belt behavior detection accuracy rate is improved.

The invention relates to a measuring tool and a method for measuring the radial and axial runout of a multi-wedge belt pulley, and belongs to the technical field of measuring equipment and measuring methods. The measuring tool comprises a rotating mechanism, a radial measuring mechanism and an axial measuring mechanism, wherein the rotating mechanism comprises a connecting shaft and a round handle; one end of the connecting shaft is fixedly provided with a standard component, and the other end is fixedly provided with the round handle; and the axial measuring mechanism comprises a cylindrical head shoulder screw, a runout measuring head, an axial runout measurement meter and a measurement meter fixing block. The measuring tool provided by the invention is reasonable in structure, convenient to use and accurate in measurement, greatly improves the detection efficiency, and is particularly suitable for being used in mass production.

The invention discloses a pedestrian head-shoulder detection and appearance capturing method based on main and subsidiary cameras. A main and subsidiary camera system is constructed, a self-made calibration board is utilized to calibrate the linkage between the main and subsidiary cameras, an improved ViBe algorithm is adopted to detect a moving object area in a scene, the problem of dead zones occurring in a traditional ViBe algorithm is solved, a pedestrian head-shoulder detection algorithm based on two stages is adopted to detect the head and shoulders of a pedestrian of the moving target area, and the problem that a traditional head-shoulder detection algorithm is high in false detection rate is solved. The pedestrian head-shoulder detection and appearance capturing method has the advantages that the moving pedestrian in the monitoring area can be detected in real time, and the distinct appearance of the moving pedestrian is obtained.

The invention provides a direct lap joint type phase solidifying pressure welding method without brazing filler metal stirring friction. Under the situation that a needle and brazing filler metal do not exist, the tool torsion function is intensified by lowering edge restraining of an upper plate (an aluminum product), a weld pass is arranged on the edge, a stir-welding head shoulder completely covers the edge of the upper plate, the upper plate edge squashing deformation and widening deformation are maximized, membrane bursting and pressing are achieved in a priority mode, the torsion function of the surface of a lower plate (steel) becomes large when the torsion deformation at the edge of the aluminum product becomes large, and membrane bursting and weld pass widening are facilitated.

The invention relates to a tapping device for a tapping and forming integrated production line, which is characterized by comprising a frame, wherein the frame is provided with a tapper; two sides of the tapper are provided with a positioning correction mechanism respectively; a working bedplate of each positioning correction mechanism is provided with two vertical guide posts; a pressing plate, a backing plate and a stripper plate are arranged on the guide posts in a sliding way; two sides of the pressing plate are connected with piston rods of cylinders; the cylinder bodies of the cylinders are fixedly connected with lateral baffles; two positioning needles are fixedly connected with the backing plate, pass through the stripper plate and are in sliding fit with the stripper plate; two hex socket head shoulder screws pass through the pressing plate and the backing plate to fixedly connect the stripper plate; and springs pressed between the backing plate and the stripper plate are sleeved on the hex socket head shoulder screws. In the tapping device for the tapping and forming integrated production line, the two sides of the tapper are provided with the positioning correction mechanism respectively, and in the tapping process, the linearity of materials is corrected, the product quality is improved and the production cost is reduced.