34results about How to "Accurate Gesture Recognition" patented technology

The embodiment of the invention provides electronic equipment and a method for carrying out projection display by utilizing the electronic equipment. The electronic equipment comprises a luminous source and a lens, wherein the luminous source comprises at least one visible light luminous source and at least one infrared luminous source; the visible light luminous source is used for emitting visible light; the infrared luminous source is used for emitting infrared light; and the lens is used for projecting out the visible light subjected to blending color mixing processing and the infrared light emitted by the infrared luminous source, so that a first projection image is protected and displayed in a projection region, wherein the visible light and the infrared light are respectively emitted into projection regions corresponding to each pixel of the first projection image. The visible light and the infrared light are respectively emitted into the projection regions corresponding to each pixel of the projection image transmitted by the electronic equipment provided by the invention, so the position of the projection region can be conveniently determined through infrared light rays reflected in the projection region, and the gesture recognition is accurately carried out.

The invention provides a gesture recognition method and system and relates to the technical field of computer vision. The gesture recognition method includes: performing multidirectional gradient extraction on an input image to form multiple different gradient images; recognizing the gradient images on the basis of a convolutional neural network pyramid to acquire an initial result set; combining the initial result set obtained by recognition to obtain a final recognition result of a gesture. Further, due to on the basis of the multidirectional gradient images describing the gesture image, ideas of a neural network and evidence theory combined recognition are integrated, so that gesture recognition accuracy is improved fully, and good recognition effect can be realized even if outside environment is similar to skin color scene and image quality is blurry.

The invention discloses a gesture identification method and system of a robot, and the robot. The method comprises steps of pre-acquiring pictures which contain different gestures and pictures which do not contain gestures so as to obtain a sample picture set; according to the sample picture set, making a detection sample set and a filtering sample set; according to the detection sample set, carrying out training so as to obtain an adaboost cascading gesture detector; according to filtering sample set, carrying out training so as to obtain a gesture identification convolution neural network; and identifying the acquired gesture pictures through the adaboost cascading gesture detector so as to obtain gesture identification results, and filtering the gesture identification results through the gesture identification convolution neural network so as to obtain a correct gesture identification result. According to the invention, by filtering the gesture identification results through the gesture identification convolution neural network, a gesture can be precisely identified under the complex background.

The invention discloses a virtual reality interaction system and method. The virtual reality interaction system comprises a smartwatch, a host machine and head equipment, wherein the smartwatch is used for acquiring user interaction information of a wearer; the host machine is in communication connection with the smartwatch and is used for receiving the user interaction information, analyzing and identifying the received user interaction information, taking the analyzed and identified user interaction information as interaction input information, processing the interaction input information and then outputting corresponding audio signals, video signals and vibrating signals; and the head equipment is in communication connection with the host machine and used for outputting video signals and audio signals. The virtual reality interaction system has the following beneficial effects that the immersive virtual reality experience is realized, and a visually immersive feeling is obtained; a rich interaction mode is realized; and vibration feedback is achieved, and the interaction experience is greatly improved.

The invention discloses a gesture-based photographing control method and device. The method comprises the following steps: identifying a gesture action of a user by use of an ultrasonic sensor in a mobile terminal; determining a shape indicated by the gesture action; and controlling a camera to photograph an image corresponding to the shape. By adopting the gesture-based photographing control method and device disclosed by the invention, the problems of the existing vision-based gesture identification method of low identification precision and poor user experience are solved, the gesture action is accurately identified, a photographing function is achieved according to the identified gesture action, and an effect of enhancing the interest and the user experience is achieved.

The invention belongs to the field of intelligent driving, and relates to a method and system for optimizing the human-computer interaction interface of an intelligent cabin based on three-branch decision. The method comprises the following steps: collecting gesture video in the cabin, preprocessing, and obtaining gesture image; segmentation of gesture image and background to obtain gesture regionimage; for multi-granularity expression, the multi-granularity feature of gesture region image being extracted by a convolution neural network; from coarse granularity to fine granularity, the imageclassification of each granularity gesture region to the conditional probability of each kind of gesture region being calculated, and the gesture recognition being accomplished sequentially by using three decision-making branches; the recognized gestures being semantically converted, and the human-computer interaction interface operating according to the result of semantic conversion. The best granularity is obtained by weighted summation, and the best granularity is used as the finest granularity. The invention not only can more accurately recognize gestures in the cockpit and execute gesturecommands, but also can reduce the interaction time of the cockpit man-machine interaction interface and provide more comfortable interaction experience for users.

The embodiment of the invention discloses a gesture recognition method, intelligent equipment and a storage medium, and belongs to the technical field of human-computer interaction. The method comprises the steps of obtaining acceleration and angular velocity data of a to-be-recognized gesture, and performing filtering processing; calculating an acceleration resultant force time sequence accordingto the acceleration data; performing dimension unified processing on the acceleration resultant force time sequence and the angular velocity time sequence; according to a preset rule, binning the acceleration resultant force time sequence and the angular velocity time sequence after the dimension unified processing; and matching the acceleration resultant force and the angular velocity time sequence after binning with a pre-stored gesture template library, and determining the type of the gesture to be recognized. According to the embodiment of the invention, under the condition that the costof the sensor is not increased, the difference of the gesture data in space can be effectively regulated, and the difference caused by speed change or force change is reduced, so that efficient and accurate gesture recognition is realized. The gesture recognition method and the gesture recognition device can be widely applied to intelligent equipment with high user interaction experience.

A method and apparatus for gesture is disclosed. The gesture recognition method comprises the following steps: acquiring a plurality of gesture images with the same preset gesture type and different gesture angles, and merging the plurality of gesture images into a multi-channel image; the multi-channel images being input into a preset convolution neural network, and the preset non-linear featuresare determined. The preset non-linear features include the 2D spatial correlation features of the images and the correlation features of various pieces of representation information of gesture objects between different channels of the images. A preselection box is generated on the feature map, and the preselection box is used to predict the position of the gesture object in the image, and the gesture recognition result is obtained according to the preset non-linear features. The invention can avoid the influence of the external environment change or the gesture posture change on the gesture recognition, and can accurately recognize the gesture no matter how the gesture posture changes, how the gesture is shielded or how the illumination condition changes, so as to improve the accuracy ofthe recognition.

The invention relates to a gesture recognition method and device, an electronic device and a storage medium, and the method comprises the steps: obtaining a monocular to-be-recognized image in a videostream; recognizing the monocular to-be-recognized image of the current frame based on a completely trained detection network to obtain a human hand bounding box of the monocular to-be-recognized image; obtaining a current static gesture result according to the human hand bounding box based on a completely trained classification network, and storing the current static gesture result in a first result set; and according to the first result set, obtaining a second result set of continuous frames in the monocular to-be-identified image, obtaining a stable static gesture result including categoryinformation according to the second result set, and obtaining a dynamic gesture result. By means of the gesture recognition method and device, the problem that gesture recognition is low in accuracyis solved, and high-speed and accurate gesture recognition is achieved.

The invention provides a finger interaction recognition system based on MEMS laser scanning. The system comprises a laser light source module which is used for emitting the RGB three-color laser; a galvanometer module which is used for receiving the laser emitted by the laser light source module and scanning the projection pictures line by line; a laser receiving module used for receiving the laser reflected during laser imaging; a depth image acquisition module which is used for acquiring a depth image with the same size as the projection pictures based on a TOF algorithm; a fingertip detection module which is used for obtaining the position and depth of a fingertip by using a deep learning algorithm; a hand image segmentation module which is used for segmenting a hand image from the depth image and determining the fingertip position; a finger action recognition module which is used for performing finger action recognition based on a neural network and a support vector machine classifier; and a finger interaction module which is used for operating and controlling the content change of the projection pictures based on the determined fingertip position. The system is small in size,is capable of being operated by hands, is convenient to carry, can realize the accurate fingertip positioning and gesture recognition, and can operate and control the projection content in real time through hands.

The invention discloses sports underwear with the function of recognizing and evaluating a human trunk exercise posture. The sports underwear comprises a sports underwear body and an exercise posture recognition and evaluation module, wherein the exercise posture recognition and evaluation module comprises a shell, a power source, an MCU, an exercise posture recognition device, an exercise evaluation device, a memorizer and a warning device. The power source supplies power and is arranged in the shell, the MCU is arranged in the shell and located above the power source, and the warning device is arranged in the shell and located on one side of the MCU which is connected with the memorizer. According to the sports underwear, the exercise posture recognition and evaluation module is fixed on the sports underwear, the correction of an initial position is simple, and the trunk exercise posture of a user is recognized accurately; the exercise posture recognition and evaluation module has the function of measuring the strength of a part of muscles of the trunk, when the user completes the physical fitness exercise, muscle load is judged by measuring the current of the muscle, if the muscle load of the user is not up to the standard, an alarm is generated to prompt the user to increase the amount of exercise or adjust the posture, and thus a better exercise effect is achieved.

The invention discloses an unmanned aerial vehicle control method based on gesture recognition. The method comprises the following steps of (1) presetting a gesture recognition remote control unmannedaerial vehicle, (2) acquiring target gesture 3D point cloud data by using a first radar chip, (3) regenerating a target gesture image, (4) providing feature information of the target gesture image, (5) obtaining a target gesture action, and (6) comparing the target gesture action with the gesture sample so as to recognize a new gesture instruction, and (7) carrying out flight action control on the unmanned aerial vehicle body or carrying out shooting action control on the camera according to the new gesture instruction. The invention further discloses the gesture recognition remote control unmanned aerial vehicle for implementing the method. The unmanned aerial vehicle control method is relatively far in control range, relatively strong in obstacle avoidance function and suitable for long-distance flight control and shooting. According to the gesture recognition remote control unmanned aerial vehicle, the gesture action is recognized through the radar chip, and compared with existinggesture visual recognition, the gesture recognition distance, speed and precision are improved.

The invention discloses an automatic coupling assembling method based on machine vision, which comprises the following steps: left half couplings are conveyed, so that the left half couplings with upward notches are normally conveyed to the lower part of an elastic body assembling platform and aligned and assembled with elastic bodies to form coupling suites; right half couplings in different directions are conveyed, so that notches of the right half couplings located on a second assembling platform face the left side; the received elastic bodies are conveyed into assembly holes, and the elastic bodies are pressed into the left half couplings with upward notches through assembly pressing rods; the first assembling platform receives the coupling suites, and the second assembling platform receives the right half couplings; and a camera group is used for identifying the positions and orientations of the left half couplings, the elastic bodies and the right half couplings, and transmitting identification results to a left side platform, a right side platform and left and right assembly platforms.

The invention discloses a gesture recognition device and method, and the device comprises a grabbing recognition mechanism which comprises a first arm which is configured to move along with an index finger; the second arm is hinged to the first arm and is configured to move along with the thumb; the offset detection mechanism is used for detecting the movement offset of the first arm relative to the second arm; and the processing module is in communication connection with the offset detection mechanism and is used for receiving the movement offset sent by the offset detection mechanism and calculating the opening degree of the first arm and the second arm according to the movement offset. The invention discloses a gesture recognition device. According to the human engineering principle, when fingers execute grabbing actions, finger postures can be changed, the arms capable of moving along with different fingers are arranged, the opening degree between the fingers is reflected, finger actions can be accurately recognized, the influence of accumulative errors of a sensor is avoided, and the detection accuracy is higher.

The invention discloses an automatic coupling assembly line based on machine vision. The assembly line comprises a left side platform, a right side platform, an elastomer assembling table, a left-right assembling platform and a camera group, the left side platform is used for conveying left half couplings, enabling the left half couplings with upward gaps to be normally conveyed to the lower part of the elastomer assembling table, and aligning and assembling with an elastomer to form a coupling suite, the right side platform is used for conveying the right half couplings in different directions so that gaps of the right half couplings located on a second assembling platform face the left side, the elastomer assembling table receives the elastomer and conveys the received elastomer into an assembling hole, the elastomer is pressed into the left half couplings with the upward gaps through an assembling pressing rod, the left-right assembling platform comprises a first assembling platform and the second assembling platform, the first assembling platform is used for receiving the coupling suite, the second assembling platform is used for receiving the right half couplings, and the camera group is used for identifying the positions and directions of the left half couplings, the elastomer and the right half couplings and conveying identification results to the left side platform, the right side platform and the left-right assembling platform.

The invention relates to the field of direct water dispensers, in particular to an AI interactive direct water dispenser based on the Internet of Things, which comprises a water accumulation structure, a direct water dispenser body, an interactive screen, a storage structure, a processing structure, a storage structure, an information acquisition structure and a sealing structure, a sealing structure used for sealing is arranged at the top end of the direct water dispenser body, a storage structure used for storing drinking water is clamped with the sealing structure, and therefore the drinking water can be conveniently stored through the storage structure, the direct water dispenser can be suitable for different environments, and meanwhile the drinking water can be stored and taken more conveniently and rapidly. The bottom end of the direct water dispenser body used for purifying a water source is provided with the water accumulation structure used for accommodating water flow, and the bottom end of the direct water dispenser body is provided with the accommodating structure which is communicated with the water accumulation structure and used for accommodating sweeps, so that thetea water sweeps can be conveniently accommodated through the accommodating structure, and water drinking is more convenient and faster.

The invention discloses a gesture recognition method and device based on multi-point TOF, and belongs to the technical field of information processing. According to the main technical scheme, the method comprises the following steps: starting an infrared time-of-flight TOF sensor; recording an entering moment when a hand to be measured enters the TOF sensor and an exiting moment when the hand to be measured exits the sensor, and calculating a signal phase calculation moment according to the entering moment and the exiting moment; and calculating a motion deviation angle of the hand to be measured relative to the central coordinate of the plane where the TOF sensing detection device is located according to the signal phase calculation moment, and determining the motion direction of the handto be measured according to the motion deviation angle and the signal phase calculation moment.

The invention discloses a multi-point touch gesture recognition method and a mistaken touch prevention method, and the gesture recognition method comprises the steps: defining a canvas touch management mode, and dividing canvas touch into gesture touch and drawing touch; defining a state conversion process in a finite state machine (FSM) in a canvas mode; judging the touch number and the touch type according to the touch area and the touch pressure intensity; adding an ID field to each touch point, and tracking the change of the touch state; eliminating the touch jitter by moving the detection threshold according to the change of the tracking touch state; in combination with the set ID, carrying out touch grouping and realizing multi-point touch gesture recognition. According to the invention, drawing with multiple single fingers and eraser operation with more than three fingers can be supported at the same time, and the situation that the system can only work to identify single touch operation is broken through; moreover, through touch grouping, simultaneous single-finger drawing of multiple persons and simultaneous erasing operation of more than three fingers of multiple persons can be realized.

The invention discloses a data glove gesture recognition algorithm based on mathematical statistics, and relates to the technical field of artificial intelligence. The method comprises the steps thata sensor.getData () method is used for a worn data glove, the same gesture is divided into a natural state and a stretching state to collect sensor angle values respectively, and the values are storedin two documents respectively; numerical values in the text document are read and stored in arrays a1 and b1, a2 and b2,..., a20 and b20, ai = (ai, 1, ai, 2, ai, 3,..., ai, 24), and bi = (bi, 1, bi,2, bi, 3,..., bi, 24). According to the method, the deviation degree and the linear relation between the randomly acquired gesture and the gesture in the template library are judged by calculating thestandard deviation and the correlation coefficient. The gesture most similar to the randomly acquired gesture is found out in the template library. Through two times of screening, the screening valueprecision is gradually improved, so that gesture recognition is very accurate. The gesture recognition based on the data glove can eliminate the interference of the external environment.

The embodiment of the invention provides a gesture recognition system and method and electronic equipment. The gesture recognition system includes: an antenna array including a plurality of millimeter wave radar transceivers configured to transmit at least one set of ranging signals and receive echo signals reflected by a target object; the data acquisition unit is configured to acquire the echo signal and cache the echo signal; and the processing unit is configured to process the cached echo signal to obtain movement information of the target object, and the movement information of the target object is used as gesture information. According to the gesture recognition system provided by the embodiment of the invention, the antenna array is simple in structure, the millimeter wave radar transceiver is adopted, the structure is simple, the size is small, the cost is low, gesture recognition can be simply realized, and the recognition rate is high.