CN117095152B - Bone recognition camera for physical training evaluation and training evaluation method - Google Patents
Bone recognition camera for physical training evaluation and training evaluation method Download PDFInfo
- Publication number
- CN117095152B CN117095152B CN202311337084.3A CN202311337084A CN117095152B CN 117095152 B CN117095152 B CN 117095152B CN 202311337084 A CN202311337084 A CN 202311337084A CN 117095152 B CN117095152 B CN 117095152B
- Authority
- CN
- China
- Prior art keywords
- unit
- human body
- main controller
- training evaluation
- bone
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000012549 training Methods 0.000 title claims abstract description 146
- 238000011156 evaluation Methods 0.000 title claims abstract description 119
- 210000000988 bone and bone Anatomy 0.000 title claims abstract description 111
- 238000012545 processing Methods 0.000 claims abstract description 95
- 238000007781 pre-processing Methods 0.000 claims abstract description 28
- 230000009471 action Effects 0.000 claims abstract description 25
- 238000005259 measurement Methods 0.000 claims description 16
- 238000000034 method Methods 0.000 claims description 15
- 230000004044 response Effects 0.000 claims description 12
- 238000003860 storage Methods 0.000 claims description 12
- 230000033001 locomotion Effects 0.000 claims description 8
- 239000000725 suspension Substances 0.000 claims description 5
- 238000012360 testing method Methods 0.000 description 11
- 210000002414 leg Anatomy 0.000 description 9
- 238000005452 bending Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000006698 induction Effects 0.000 description 4
- 210000003127 knee Anatomy 0.000 description 4
- 238000011065 in-situ storage Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 210000000323 shoulder joint Anatomy 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 210000000629 knee joint Anatomy 0.000 description 2
- 235000013490 limbo Nutrition 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000035484 reaction time Effects 0.000 description 2
- 238000000926 separation method Methods 0.000 description 2
- 206010065687 Bone loss Diseases 0.000 description 1
- 241001622623 Coeliadinae Species 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000001915 proofreading effect Effects 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
- G06V10/12—Details of acquisition arrangements; Constructional details thereof
- G06V10/14—Optical characteristics of the device performing the acquisition or on the illumination arrangements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
- G06V10/12—Details of acquisition arrangements; Constructional details thereof
- G06V10/14—Optical characteristics of the device performing the acquisition or on the illumination arrangements
- G06V10/147—Details of sensors, e.g. sensor lenses
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/23—Recognition of whole body movements, e.g. for sport training
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/50—Constructional details
- H04N23/55—Optical parts specially adapted for electronic image sensors; Mounting thereof
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Signal Processing (AREA)
- General Health & Medical Sciences (AREA)
- Vascular Medicine (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses a skeleton recognition camera for physical training evaluation and a training evaluation method, which relate to the fields of national body evaluation and physical training evaluation, and comprise the following components: the system comprises a PCB main controller, a vision preprocessing unit, a ranging unit and an NPU processing unit; the PCB main controller is used for: receiving a training evaluation course coding instruction sent by an upper computer to determine a corresponding course skeleton point model; acquiring a human body video picture of physical training evaluation; the NPU processing unit is used for: determining skeleton point information of a human body according to a human body video picture; the vision preprocessing unit is used for: comparing the skeleton point information with a corresponding lesson skeleton point model, and judging whether the action of the human body reaches the standard or not; the distance measuring unit is used for determining the distance information of the human body and the bone recognition camera; the PCB main controller is used for: and transmitting the skeleton point information, the distance information and the judging result to an upper computer. The invention can accurately realize bone identification for physical training evaluation with low cost.
Description
Technical Field
The invention relates to the field of national body measurement and physical training evaluation, in particular to a bone recognition camera for physical training evaluation and a training evaluation method.
Background
With the development of national body measurement and physical training and evaluation, the physical training and evaluation have the advantages of more personnel, strict standard, wide coverage range and the like. The national sports affairs relatives are healthy, the sports training and the evaluation have definite standards on soldiers, policemen, firemen, athletes and students, the data acquisition accuracy is fair and fair, the training and the evaluation data acquisition speed are high, and the relatives are evaluated with high efficiency. The national body measurement and physical training and evaluation equipment is mainly deployed in a track and field training field and a special training field, has high requirements on environment, needs to adapt to illumination conditions of deployment at different angles, and can accurately identify national body measurement and physical training and evaluation action and gesture data; the training and evaluation method is accurate, safe and reliable, and can effectively improve the efficiency of national body test and sports subject test data acquisition.
The prior art provides a sit-up intelligent testing device, includes: the sit-up plate is provided with a knee joint propping support, and the extension plate is provided with a foot fixing support; the infrared opposite-shooting sensors are arranged at intervals in a mode of being high in front and low in back on the left side and the right side of the sit-up plate, the induction device is arranged at the rear end of the sit-up plate far away from the knee joint propping support, the infrared opposite-shooting sensors and the induction device are connected with the main control device, the main control device can preset that the infrared opposite-shooting sensors are in a working state, and therefore an infrared stroke induction zone which is suitable for the sit-up test process of a tested person is formed between the infrared opposite-shooting sensors in the working state. The scheme adopts an infrared travel induction area technology, cannot accurately judge unified standards of high, low, fat and thin, and cannot be used outdoors in all weather.
The prior art also provides a sit-up tester and a physical ability testing system, the sit-up tester comprises a sensing pad and a knee pad, the knee pad is used for being worn at the knee of a tester, a first sensor is integrated in the sensing pad, a second sensor, a counter and a timer are integrated in the knee pad, the first sensor and the second sensor are both connected with the counter through signals, the first sensor is used for acquiring the prone position action of the tester, the second sensor is used for acquiring the sitting up action of the tester, and the counter is used for calculating the number of the sit-ups completed by the tester according to the prone position action and the times of the sitting up action; this sit up tester conveniently carries and deposits, has greatly made things convenient for the condition that needs to put equipment outdoor and test, and the tester is wearing formula and the form of combining together of placing, can personally place and dress according to tester's height size, and the in-process accuracy is high. The tester is in a combined form of wearing type and placement type, has the problems of different wearing orientations and different acquired data, and is low in accuracy.
The prior art also provides an intelligent horizontal bar pull-up evaluation training system based on the skeleton recognition technology, wherein the camera is used for acquiring videos of an evaluated person in the horizontal bar evaluation process; the computing device comprises a storage module, a first identification module and a second identification module; the storage module is used for storing all frame images in the video to a collection frashv; the first recognition module is used for respectively acquiring a set seltlow1 and a set selhig1 of frame images of the lowest point and the highest point which represent the first complete pull-up motion from the set frashv; calculating selection parameters based on seltlow1 and selhig1; deleting the frame image of the first complete pull-up action from the set ofram to obtain an updated set ofram; the second recognition module is used for recognizing the frame images in the set nfraphv based on the selection parameters, and calculating the number of pull-up directions completed by the evaluated personnel. The scheme adopts a camera and computing equipment combination, belongs to a front-end and back-end combination method, and a computer carries out AI reasoning operation through a front-collected video picture and a back-end processing calculation.
The prior physical training and evaluation equipment mainly adopts infrared, laser, wearable angle sensors and front and rear end bone recognition technologies, and adopts the infrared and laser equipment to judge whether the action is standard or not accurately due to the influence of weather and light and invisible actions; the wearable angle sensor has different wearing directions, different acquired data and low accuracy; the front-end and rear-end bone recognition technology adopts a front-end camera to collect images and a rear-end computer to perform reasoning operation, has high equipment cost and severe requirements on use and deployment environments, and is not beneficial to large-scale popularization.
In summary, how to accurately realize physical training and evaluation of bone recognition with low cost becomes a problem to be solved urgently at present.
Disclosure of Invention
Based on the above, the embodiment of the invention provides the bone identification camera for the physical training evaluation and the training evaluation method, which can accurately realize bone identification for the physical training evaluation with low cost.
In order to achieve the above object, the embodiment of the present invention provides the following solutions:
a bone recognition camera for physical training assessment, comprising: the system comprises a PCB main controller, a vision preprocessing unit, a ranging unit and an NPU processing unit;
the PCB main controller is internally provided with the vision pretreatment unit; the PCB main controller is respectively connected with the vision preprocessing unit, the ranging unit and the NPU processing unit; the PCB main controller is used for connecting with an upper computer;
the PCB main controller is used for:
receiving a training evaluation course coding instruction sent by the upper computer, determining a corresponding course skeleton point model according to the training evaluation course coding instruction, and sending the course skeleton point model to the vision preprocessing unit;
acquiring a human body video picture of physical training evaluation, and transmitting the human body video picture to the NPU processing unit;
the NPU processing unit is used for:
determining skeleton point information of a human body according to the human body video picture, and sending the skeleton point information to the vision preprocessing unit and the PCB main controller;
the vision preprocessing unit is used for:
comparing the skeleton point information with a corresponding lesson skeleton point model, judging whether the action of a human body reaches the standard to obtain a judging result, and sending the judging result to a PCB main controller;
the distance measuring unit is used for determining the distance information of the human body and the bone recognition camera and sending the distance information to the PCB main controller; the distance information is used for determining whether the bone point information is complete;
the PCB main controller is also used for:
and sending the bone point information, the distance information and the judgment result to the upper computer.
Optionally, the bone recognition camera for physical training evaluation further comprises: the image processing device comprises a control unit, an image processing unit, a first lens and a second lens;
the control unit is connected with the first lens; the control unit is connected with the PCB main controller through the image processing unit; the second lens is connected with the ranging unit;
the control unit is used for controlling the first lens to collect the on-site picture of the human body; the image processing unit is used for processing the field picture to obtain a human body video picture;
the distance measuring unit is used for sending out light pulses through the second lens, receiving scattered light reflected by a human body through the second lens, calculating a calculated distance value between the human body and the second lens according to the light pulses and the scattered light, and taking the calculated distance value as distance information of a human body and a bone recognition camera.
Optionally, the bone recognition camera for physical training evaluation further comprises a coding unit;
the coding unit is respectively connected with the control unit and the PCB main controller; the encoding unit is used for encoding the field picture to obtain an encoded video and transmitting the encoded video to the PCB main controller;
and the PCB main controller is also used for sending the coded video to the upper computer.
Optionally, the bone recognition camera for physical training evaluation further comprises: the audio acquisition unit and the voice broadcasting unit;
the audio acquisition unit and the voice broadcasting unit are connected with the PCB main controller;
the audio acquisition unit is used for acquiring audio content of the environment where the human body is located and sending the audio content to the upper computer through the PCB main controller;
the PCB main controller is also used for sending voice reminding information to the voice broadcasting unit when the distance information exceeds a set distance range;
the voice broadcasting unit is used for broadcasting the voice reminding information.
Optionally, the bone recognition camera for physical training evaluation further comprises: a DSP digital signal processing unit and a network interface;
the main controller is connected with the network interface through the DSP digital signal processing unit; the network interface is used for connecting the upper computer.
Optionally, the bone recognition camera for physical training evaluation further comprises: ARM processing unit;
the ARM processing unit is connected with the PCB main controller; the ARM processing unit is used for setting lens parameters for the first lens and the second lens according to the training evaluation object coding instruction.
Optionally, the bone recognition camera for physical training evaluation further comprises: a memory unit;
the storage memory unit is connected with the PCB main controller; the storage memory unit is used for storing various lesson and mesh skeleton point models for the purpose of physical training evaluation; the physical training assessment subjects include: pull-up, arm-curl suspension, sit-up, push-up, arm-curl extension, parallel bar forward movement, seat forward, extension jump, longitudinal jump, closed-eye single-foot standing, S-shape running and middle-long running.
Optionally, the PCB main controller is further configured to:
if the judging result is up to standard, adding 1 to the effective number of the physical training evaluation, and if the judging result is not up to standard, adding 1 to the ineffective number of the physical training evaluation;
transmitting training evaluation data to the upper computer; the training evaluation data includes: the method comprises the steps of training evaluation object names, training evaluation dates, training evaluation time, personnel names, training evaluation effective times, training evaluation ineffective times and training evaluation time.
Optionally, the upper computer is further configured to send a request to the PCB host controller; the PCB main controller is also used for sending a corresponding response to the upper computer after receiving the request; the upper computer is also used for determining that the connection state is currently in the connection state if corresponding response is received after the request is sent.
The invention also provides a training evaluation method which is realized by adopting the bone recognition camera for physical training evaluation; the training evaluation method comprises the following steps:
receiving a training evaluation class coding instruction sent by an upper computer;
determining a corresponding course skeleton point model according to the training evaluation course coding instruction;
acquiring a human body video picture of physical training evaluation, and sending the human body video picture to an NPU processing unit;
receiving skeleton point information of a human body; the skeleton point information is determined by the NPU processing unit according to the human body video picture;
receiving a judgment result of the vision pretreatment unit; the judging result is that the vision preprocessing unit compares the skeleton point information with a corresponding lesson skeleton point model, and judges whether the action of the human body reaches the standard or not;
receiving distance information of a human body and a bone recognition camera, which is determined by a distance measurement unit; the distance information is used for determining whether the bone point information is complete;
and sending the bone point information, the distance information and the judgment result to the upper computer.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the embodiment of the invention provides a bone recognition camera for physical training evaluation and a training evaluation method, wherein the camera comprises the following components: the PCB main controller is used for being connected with an upper computer, the camera is used for national body survey and sports lessons, the problems of weather, light, invisible action standard and the like of traditional infrared, laser and wearable angle sensors are solved, the camera can be deployed at any position and any direction of a standard training place, bone information points can be obtained at the same time, whether national body survey and sports lessons actions are standard or not can be accurately judged, and bone recognition for physical training and evaluation can be accurately realized; the problems of high equipment cost, complex deployment environment and the like caused by the bone identification separation of the front end and the rear end of the traditional AI bones are solved, an integrated bone identification camera is adopted, and a PCB main controller, a vision preprocessing unit, a ranging unit and an NPU processing unit are embedded, so that the acquisition efficiency is improved, the equipment cost is saved, and the installation cost is simplified.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic structural diagram of a bone recognition camera for physical training evaluation according to an embodiment of the present invention;
FIG. 2 is a workflow diagram of a training assessment method implemented using a bone recognition camera according to an embodiment of the present invention;
FIG. 3 is a diagram showing a 21-point skeleton point information distribution diagram obtained by an NPU processing unit according to an embodiment of the present invention;
fig. 4 is an exemplary diagram of an interface displayed by a host computer according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be described below with reference to the accompanying drawings in the embodiments of the present invention.
The invention aims to provide a bone recognition camera for physical training evaluation and a training evaluation method, which are used for accurately realizing bone recognition for physical training evaluation with low cost by integrating a PCB main controller, a vision preprocessing unit, a ranging unit and an NPU processing unit into the bone recognition camera.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
Referring to fig. 1, the bone recognition camera for sports training evaluation of the present embodiment includes: the system comprises a PCB main controller, a vision preprocessing unit, a ranging unit and a NPU (Neural Processing Unit) processing unit. The NPU processing unit is a neural network processor.
The PCB main controller is internally provided with the vision pretreatment unit; the PCB main controller is respectively connected with the vision preprocessing unit, the ranging unit and the NPU processing unit; the PCB main controller is used for being connected with an upper computer.
The PCB main controller is used for: receiving a training evaluation course coding instruction sent by the upper computer, determining a corresponding course skeleton point model according to the training evaluation course coding instruction, and sending the course skeleton point model to the vision preprocessing unit; and acquiring a human body video picture of the physical training evaluation, and sending the human body video picture to the NPU processing unit.
The NPU processing unit is internally provided with an inference algorithm for: and determining skeleton point information of a human body according to the human body video picture, and sending the skeleton point information to the vision preprocessing unit and the PCB main controller.
The vision preprocessing unit is used for: and comparing the skeleton point information with a corresponding lesson skeleton point model, judging whether the action of the human body reaches the standard or not to obtain a judging result, and sending the judging result to a PCB main controller.
The distance measuring unit is used for determining the distance information of the human body and the bone recognition camera and sending the distance information to the PCB main controller; the distance information is used to determine whether the skeletal point information is complete.
The PCB main controller is also used for: and sending the bone point information, the distance information and the judgment result to the upper computer.
In one example, the NPU processing unit is built with a 2s-AGCN algorithm for reasoning about skeletal point information of the human body. The reasoning formula of the 2s-AGCN algorithm is as follows:
;
wherein f out Is skeletal point information; w (W) k Is the weight; a is that k Is an adjacency matrix; b (B) k As a trainable N x N matrix, it can not only indicate whether there is a link between two skeletal points, but alsoRepresenting the strength of the connection; c (C) k For one unique graph learned for each sample, a gaussian embedding function is used to capture the similarity between two skeletal points, the resulting probability of 0-1 is processed by a Softmax function; k (K) v The number of subgraphs is generally 3 (corresponding to rest, centripetal and centrifugal respectively); k represents the number of the subgraph; f (f) in Is an input human body video picture.
Generating C using Gaussian embedding function k The specific calculation formula is as follows:
;
W θk the weight of the θ branch;representation->The weight of the branch; t represents the transpose.
In one example, still referring to fig. 1, the bone recognition camera for physical training assessment further includes: DSP (Digital Signal Processing) digital signal processing units and network interfaces. The network interface may be a PHY (Physical Layer) network interface.
The main controller is connected with the network interface through the DSP digital signal processing unit; the network interface is used for connecting the upper computer.
And the PCB main controller converts the skeleton point information, the distance information and the judging result into CRC16 instructions and sends the CRC16 instructions to the upper computer through the DSP digital signal processing unit and the network interface.
In one example, still referring to fig. 1, the bone recognition camera for physical training assessment further includes: ARM processing unit. And the ARM processing unit is connected with the PCB main controller.
The ARM processing unit is provided with an embedded control system, and can be used for configuring parameters of the bone recognition camera and secondarily developing embedded functions. For example, the ARM processing unit may be configured to set shot parameters for the first shot and the second shot according to a training evaluation objective code instruction.
In one example, still referring to fig. 1, the bone recognition camera for physical training assessment further includes: a control unit, an image processing unit (Image Signal Processor, ISP), a first lens and a second lens.
The control unit is connected with the first lens; the control unit is connected with the PCB main controller through the image processing unit; the second lens is connected with the ranging unit.
The control unit is used for controlling the first lens to collect the on-site picture of the human body; the image processing unit is used for processing the field picture to obtain a human body video picture.
The distance measuring unit is a TOF (Time of Flight) distance measuring unit and is used for sending out light pulses through the second lens, receiving scattered light reflected by a human body through the second lens, calculating a calculated distance value between the human body and the second lens according to the light pulses and the scattered light, and taking the calculated distance value as distance information of a human body and a bone recognition camera.
The TOF ranging unit is specifically used for calculating the calculated distance value between the human body and the second lens: after exiting through the light pulse transmitting light path, the light pulse reaches the surface of a measured object (such as a human body) through a second lens and is scattered in all directions, the second lens receives part of scattered light, part of the scattered light enters a photoelectric device of a TOF ranging unit through a receiving light path of the TOF ranging unit, the photoelectric device is converted into photocurrent, the photocurrent is transmitted to an echo signal processing circuit, and after being converted into CRC16 instructions, distance information is output to an upper computer through a DSP digital signal processing unit and a network interface.
In one example, still referring to fig. 1, the bone recognition camera for physical training assessment further includes an encoding unit. The coding unit is respectively connected with the control unit and the PCB main controller; the encoding unit is used for encoding the field picture to obtain an encoded video, and transmitting the encoded video to the PCB main controller. And the PCB main controller is also used for sending the coded video to the upper computer.
Specifically, the control unit is a CCD/CMOS control unit; the coding unit is an NTSC/PAL coding unit. The NTSC/PAL coding unit is connected with the CCD/CMOS control unit to obtain the lens picture, the NTSC format or PLA format video coding is adopted, the DSP digital signal processing unit is connected with the PCB main controller to be converted into digital signals, and the digital signals are connected with the upper computer through the network interface.
The CCD/CMOS control unit is connected with the first lens and is used for generating an optical image by the light sensor, and after the optical image is processed by the image processing unit, the optical image is sent to the NPU processing unit by the PCB main controller for processing and identification; when the digital signal is output, the digital signal is converted into a digital signal through the NTSC/PAL coding unit and the digital signal is connected with an upper computer through a network interface.
In one example, still referring to fig. 1, the bone recognition camera for physical training assessment further includes: the audio acquisition unit and the voice broadcasting unit.
The audio acquisition unit and the voice broadcasting unit are connected with the PCB main controller.
The audio acquisition unit is used for acquiring audio content of the environment where the human body is located and sending the audio content to the upper computer through the PCB main controller. The upper computer stores the audio content in the audio and video storage device.
The PCB main controller is also used for sending voice reminding information to the voice broadcasting unit when the distance information exceeds a set distance range (for example, 0.5-4 meters).
The voice broadcasting unit is used for broadcasting the voice reminding information.
The distance information exceeds the set distance range, the distance cannot be measured, the distance measuring unit mainly judges whether the distance between a person and the second lens is too short or too long, the whole body cannot be identified due to the fact that the distance is too short, the action cannot be judged, the too long identification rate can be reduced, bone loss can occur, therefore, the optimal position is obtained through the distance measuring unit, and voice output reminding is conducted through the voice broadcasting unit. For example, the voice broadcasting unit sends out voice reminding information of 'too close' less than 0.5 meter to remind people to get far away from the second lens; more than 4 meters, the voice broadcasting unit sends out voice reminding information of 'far' to remind people to get close to the second lens; when the distance is in the range of 0.5-4 meters, the complete bone point information can be obtained.
It should be noted that, the voice broadcasting unit is further configured to broadcast a voice playing request sent by the upper computer and play a voice content required by training and checking.
In one example, the bone recognition camera for physical training assessment further comprises: the memory cell is stored. The storage memory unit is connected with the PCB main controller; the storage memory unit is used for storing a model library, wherein the model library comprises various lesson and mesh skeleton point models for the purpose of physical training evaluation; the physical training assessment subjects include: pull-up, arm-bending suspension, sit-up, push-up, arm extension, parallel bar forward movement, closed eye single foot standing, seat forward flexion, in-situ longitudinal jump, reaction time test, step, high leg lifting, deep squat, stride, straight bow squat, shoulder joint flexibility, straight leg active upper table, trunk stabilization push-up, rotational stability, body posture, high leg lifting, S-shaped running, middle and long running and the like.
In one example, the PCB main controller is further to: if the judging result is up to standard, adding 1 to the effective number of the physical training evaluation, and if the judging result is not up to standard, adding 1 to the ineffective number of the physical training evaluation; and sending the training evaluation data to the upper computer. The training evaluation data includes: the method comprises the steps of training evaluation object names, training evaluation dates, training evaluation time, personnel names, training evaluation effective times, training evaluation ineffective times and training evaluation time.
In one example, the host computer is further configured to send a request to the PCB host controller; the PCB main controller is also used for sending a corresponding response to the upper computer after receiving the request; the upper computer is also used for determining that the connection state is currently in the connection state if corresponding response is received after the request is sent.
The ARM processing unit realizes secondary development of the bone recognition camera, the upper computer is connected with the bone recognition camera, the ARM processing unit has the functions and parameters of the bone recognition camera can be edited and burnt, and the functions and parameters of the bone recognition camera can be adjusted, such as: the default identification is pull-up, the corresponding model of the lesson skeleton point corresponding to pull-up of the storage memory unit is read, if the model needs to be changed, the model needs to be domesticated again, the corresponding algorithm mark needs to be burnt into the ARM processing unit, and otherwise, the model stored by the storage memory unit cannot be matched with the mark in the ARM processing unit.
In addition, the bone recognition camera also comprises other matching interfaces to realize corresponding functions. Still referring to fig. 1, for example, the bone recognition camera further includes: USB, I/O unit, audio interface and power interface, etc.
Taking an audio acquisition unit as an example, the audio acquisition unit is connected with a PCB controller, and is externally connected with a loudspeaker and a pickup through an audio interface, so that the audio content of an acquisition site is adopted, and a voice command is broadcasted.
When the digital signal processing device is used, the bone recognition camera is electrified and networked, the upper computer is connected, the upper computer is used for acquiring a real-time picture of the bone recognition camera in real time, acquiring bone point information, a 2s-AGCN reasoning algorithm embedded in the NPU processing unit is used for processing a video picture acquired by the DSP digital signal processing unit in real time, extracting human body information in the picture into bone point information, and then converting the bone point information into CRC16 instructions to be output through the DSP digital signal processing unit.
When the intelligent human body digital signal processing device is used, the upper computer is connected with the bone recognition camera to acquire a real-time video image, the camera is within the range of 0.5-4 meters from a human body, after the human body enters the image range, the upper computer selects an evaluation object, and after identification information is identified through hand lifting schematic, counting, timing and action standard comparison are started, and each time of qualified actions are added, the bone recognition camera automatically accumulates once, the effective times are increased by 1, the actions are unqualified, the ineffective times are increased by 1, and the data information of the camera is output through the DSP digital signal processing unit.
The working flow of the training evaluation method realized by adopting the bone recognition camera is shown in fig. 2, the distribution of the bone point information of 21 points obtained by the NPU processing unit is shown in fig. 3, the interface displayed by the upper computer is shown in fig. 4, and the information such as the bone point information, the name, the effective times, the ineffective times, the date and the time length of 21 points which are evaluated in pull-up mode is shown in fig. 4.
The bone recognition camera of the embodiment is an AI bone recognition camera, and has the following advantages:
1) The AI skeleton recognition camera is used for national body survey and sports lessons, has the advantages that the problems that the traditional infrared, laser and wearable angle sensor is affected by weather and light, cannot accurately judge whether the motion is standard or not due to invisible motions are solved, the AI skeleton recognition camera is adopted, the AI skeleton recognition camera can be deployed at any position and any direction of a standard training place, meanwhile, the AI skeleton recognition camera is used for collecting visible AI skeletons, and real-time images of the national body survey and sports lessons can be accurately judged.
2) The AI skeleton recognition camera is used for national body survey and sports lessons, and has the advantages of solving the problems of high equipment cost, complex deployment environment and the like caused by the separation of the front end skeleton and the rear end skeleton of the traditional AI skeleton, adopting the integrated AI skeleton recognition camera, embedding a PCB main controller, a vision preprocessing unit, a TOF ranging unit, an NPU processing unit and an ARM processing unit, improving the acquisition efficiency, saving the equipment cost and simplifying the installation cost.
3) The AI skeleton recognition camera is used for national body measurement and sports lessons, has the advantages of improving operation and reasoning efficiency, setting the skeleton recognition reasoning inside the camera through the PCB main controller and the vision preprocessing unit, starting timing, counting, standard checking and generating a human serial number, finishing the timing inside the camera, only feeding back training and evaluating data to the upper computer, saving data transmission network resources and feedback time, reducing the delay problem of the return values of video pictures and the upper computer, and greatly improving the data accuracy and the equipment operation efficiency.
4) The AI skeleton recognition camera is used for national body measurement and sports lessons, and has the advantages that the front-end camera can autonomously organize national body measurement and sports lessons training and evaluation, and comprises the lessons such as pull-up, arm bending suspension, sit-up, push-up, double-arm bending extension, parallel bar forward movement, closed-eye single-foot standing, seat forward bending, in-situ longitudinal jump, reaction test, steps, high leg lifting, deep squatting, stride, straight bow-step squatting, shoulder joint flexibility, straight leg active upper stage, trunk stable push-up, rotation stability, body type, high leg lifting, S-shaped running, middle-long running and the like, and meanwhile, the training of sports and the lesson model library can be continuously increased and expanded.
The invention also provides a training evaluation method which is realized by adopting the bone recognition camera for physical training evaluation; the training evaluation method comprises the following steps:
(1) And receiving a training evaluation program coding instruction sent by the upper computer.
(2) And determining a corresponding course skeleton point model according to the training evaluation course coding instruction.
(3) And acquiring a human body video picture of the physical training evaluation, and sending the human body video picture to an NPU processing unit.
(4) Receiving skeleton point information of a human body; the skeleton point information is determined by the NPU processing unit according to the human body video picture.
(5) Receiving a judgment result of the vision pretreatment unit; the judging result is that the vision preprocessing unit compares the skeleton point information with a corresponding lesson skeleton point model, and judges whether the action of the human body reaches the standard or not.
(6) Receiving distance information of a human body and a bone recognition camera, which is determined by a distance measurement unit; the distance information is used to determine whether the skeletal point information is complete.
(7) And sending the bone point information, the distance information and the judgment result to the upper computer.
In practical application, a specific implementation process of the training evaluation method is as follows:
step 1: the intelligent training and evaluation system is used for national body measurement and physical training and evaluation, and comprises a pull-up, arm-bending suspension, sit-up, push-up, double-arm bending, parallel bar forward movement, closed-eye single foot standing, seat body forward bending, in-situ longitudinal jump, reaction time test, steps, high leg lifting, deep squat, stride, straight bow squat, shoulder joint flexibility, straight leg active upper stage, trunk stable push-up, rotation stability, body posture, high leg lifting, S-shaped running, middle-long running and the like class, an upper computer selects class codes, is connected with an AI recognition camera through a TCP, transmits training and evaluation class code instructions, transmits requests, and receives responses including registration requests, heartbeat responses, class information requests, class information responses, skeletal point reporting requests, skeleton point reporting responses, class score requests, class score responses, violation requests and violation responses.
Step 2: when the method is used for national body measurement, physical training and class assessment, an upper computer is connected with a bone identification camera through a TCP, a DSP digital signal processing unit in the bone identification camera receives a class encoding request instruction, then a vision preprocessing unit is used for extracting bone point information of ISP pictures identified by an NPU processing unit, an adaptive graph is adopted for convolution, bone point information of 10 points, 17 points and 21 points can be extracted, after the bone identification camera receives the instruction, timing is started, video pictures display class names, dates, time, names, effective times, ineffective times, duration and other data in real time, and the method returns to the upper computer through the DSP digital signal processing unit.
Step 3: when the distance measuring unit is used for national body measurement and physical education training and evaluation, the range of the measured human body distance bone recognition camera is larger than 0.5 to 4 meters, and the distance measuring unit is connected with the DSP digital signal processing unit through the PCB main controller to actively feed back the distance information between the human body and the bone camera and is used for assisting in judging whether the bone point information fed back by the NPU processing unit is complete or not.
Step 4: when the method is used for national body measurement and physical education course training and evaluation, the upper computer is connected with the bone recognition camera through the TCP, sends a course coding request instruction and a training and evaluation starting instruction, the PCB main controller downwards reaches the NPU processing unit, identifies the bone point information of an ISP picture, judges whether the extracted bone point information is accurate with a model library at the vision preprocessing unit, returns to the upper computer real-time video picture, displays the extracted bone point information of a human body on the picture, displays training and evaluation data such as a course name, date, time, name, effective times, ineffective times, duration and the like in real time, and also returns to the upper computer general data through the DSP digital signal processing unit.
Step 5: when the method is used for national body measurement and training of the sports lesson model library, the upper computer is connected with the bone recognition camera through the TCP, the PCB main controller is connected with the vision preprocessing unit, and the upper computer can write the lesson bone recognition model library into the vision preprocessing unit for national body measurement and training of sports lesson, and the evaluation provides comparison basis.
The bone recognition camera of the embodiment is an AI bone recognition camera, a built-in reasoning algorithm and a model library are built in, the completion inside the AI bone recognition camera is realized through an ARM subsystem (comprising an NPU processing unit and an ARM processing unit), automatic acquisition, reasoning and operation are carried out, national body survey and sports class bone point information are output according to instructions, and after the upper computer sends class requests through the instructions, data such as action timing start, timing, counting, standard proofreading, figure number and the like are judged through the bone information and gestures; the training and evaluating method is accurate, safe and reliable, and can effectively improve national body tests, training of sports lessons, testing accuracy and data acquisition efficiency.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the method disclosed in the embodiment, since it corresponds to the device disclosed in the embodiment, the description is relatively simple, and the relevant points are referred to the device part description.
The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the methods of the present invention and the core ideas thereof; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.
Claims (8)
1. A bone recognition camera for physical training assessment, comprising: the system comprises a PCB main controller, a vision preprocessing unit, a ranging unit and an NPU processing unit;
the PCB main controller is internally provided with the vision pretreatment unit; the PCB main controller is respectively connected with the vision preprocessing unit, the ranging unit and the NPU processing unit; the PCB main controller is used for connecting with an upper computer;
the PCB main controller is used for:
receiving a training evaluation course coding instruction sent by the upper computer, determining a corresponding course skeleton point model according to the training evaluation course coding instruction, and sending the course skeleton point model to the vision preprocessing unit;
acquiring a human body video picture of physical training evaluation, and transmitting the human body video picture to the NPU processing unit;
the NPU processing unit is used for:
determining skeleton point information of a human body according to the human body video picture, and sending the skeleton point information to the vision preprocessing unit and the PCB main controller;
the vision preprocessing unit is used for:
comparing the skeleton point information with a corresponding lesson skeleton point model, judging whether the action of a human body reaches the standard to obtain a judging result, and sending the judging result to a PCB main controller;
the distance measuring unit is used for determining the distance information of the human body and the bone recognition camera and sending the distance information to the PCB main controller; the distance information is used for determining whether the bone point information is complete;
the PCB main controller is also used for:
transmitting the skeleton point information, the distance information and the judgment result to the upper computer;
the bone recognition camera for physical training evaluation further comprises: the image processing device comprises a control unit, an image processing unit, a first lens and a second lens;
the control unit is connected with the first lens; the control unit is connected with the PCB main controller through the image processing unit; the second lens is connected with the ranging unit;
the control unit is used for controlling the first lens to collect the on-site picture of the human body; the image processing unit is used for processing the field picture to obtain a human body video picture;
the distance measuring unit is used for sending out light pulses through the second lens, receiving scattered light reflected by a human body through the second lens, calculating a calculated distance value between the human body and the second lens according to the light pulses and the reflected light, and taking the calculated distance value as distance information of a human body and a bone recognition camera;
the bone recognition camera for physical training evaluation further comprises: a DSP digital signal processing unit and a network interface;
the main controller is connected with the network interface through the DSP digital signal processing unit; the network interface is used for connecting the upper computer;
when the digital signal processing device is used, the bone recognition camera is electrified and networked, and is connected with the upper computer, the upper computer is used for acquiring a real-time picture of the bone recognition camera in real time, acquiring bone point information, processing a video picture acquired by the DSP digital signal processing unit in real time through a 2s-AGCN reasoning algorithm embedded in the NPU processing unit, extracting human body information in the picture into bone point information, and converting the bone point information into a CRC16 instruction to be output through the DSP digital signal processing unit;
when the intelligent human body digital signal processing device is used, the upper computer is connected with the bone recognition camera to acquire a real-time video image, the camera is within the range of 0.5-4 meters from a human body, after the human body enters the image range, the upper computer selects an evaluation object, and after identification information is identified through hand lifting schematic, counting, timing and action standard comparison are started, and each time of qualified actions are added, the bone recognition camera automatically accumulates once, the effective times are increased by 1, the actions are unqualified, the ineffective times are increased by 1, and the data information of the camera is output through the DSP digital signal processing unit.
2. The bone recognition camera for physical training assessment of claim 1, further comprising: a coding unit;
the coding unit is respectively connected with the control unit and the PCB main controller; the encoding unit is used for encoding the field picture to obtain an encoded video and transmitting the encoded video to the PCB main controller;
and the PCB main controller is also used for sending the coded video to the upper computer.
3. The bone recognition camera for physical training assessment of claim 1, further comprising: the audio acquisition unit and the voice broadcasting unit;
the audio acquisition unit and the voice broadcasting unit are connected with the PCB main controller;
the audio acquisition unit is used for acquiring audio content of the environment where the human body is located and sending the audio content to the upper computer through the PCB main controller;
the PCB main controller is also used for sending voice reminding information to the voice broadcasting unit when the distance information exceeds a set distance range;
the voice broadcasting unit is used for broadcasting the voice reminding information.
4. The bone recognition camera for physical training assessment of claim 2, further comprising: ARM processing unit;
the ARM processing unit is connected with the PCB main controller; the ARM processing unit is used for setting lens parameters for the first lens and the second lens according to the training evaluation object coding instruction.
5. The bone recognition camera for physical training assessment of claim 1, further comprising: a dedicated storage memory unit;
the special storage memory unit is connected with the PCB main controller; the special storage memory unit is used for storing various lesson and mesh skeleton point models for the purpose of physical training evaluation; the physical training assessment subjects include: pull-up, arm-curl suspension, sit-up, push-up, arm-curl extension, parallel bar forward movement, seat forward flexion, standing long jump, vertical jump, closed-eye single foot standing, S-shape running and middle-long running.
6. The bone recognition camera for physical training assessment of claim 1, wherein said PCB main controller is further configured to:
if the judging result is up to standard, adding 1 to the effective number of the physical training evaluation, and if the judging result is not up to standard, adding 1 to the ineffective number of the physical training evaluation;
transmitting training evaluation data to the upper computer; the training evaluation data includes: the method comprises the steps of training evaluation object names, training evaluation dates, training evaluation time, personnel names, training evaluation effective times, training evaluation ineffective times and training evaluation time.
7. The bone recognition camera for physical training assessment of claim 1, wherein said host computer is further configured to send a request to said PCB host controller; the PCB main controller is also used for sending a corresponding response to the upper computer after receiving the request; the upper computer is also used for determining that the connection state is currently in the connection state if corresponding response is received after the request is sent.
8. A training evaluation method, characterized in that the training evaluation method is implemented by using the bone recognition camera for physical training evaluation according to any one of claims 1 to 7; the training evaluation method comprises the following steps:
receiving a training evaluation class coding instruction sent by an upper computer;
determining a corresponding course skeleton point model according to the training evaluation course coding instruction;
acquiring a human body video picture of physical training evaluation, and sending the human body video picture to an NPU processing unit;
receiving skeleton point information of a human body; the skeleton point information is determined by the NPU processing unit according to the human body video picture;
receiving a judgment result of the vision pretreatment unit; the judging result is that the vision preprocessing unit compares the skeleton point information with a corresponding lesson skeleton point model, and judges whether the action of the human body reaches the standard or not;
receiving distance information of a human body and a bone recognition camera, which is determined by a distance measurement unit; the distance information is used for determining whether the bone point information is complete;
and sending the bone point information, the distance information and the judgment result to the upper computer.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311337084.3A CN117095152B (en) | 2023-10-17 | 2023-10-17 | Bone recognition camera for physical training evaluation and training evaluation method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311337084.3A CN117095152B (en) | 2023-10-17 | 2023-10-17 | Bone recognition camera for physical training evaluation and training evaluation method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117095152A CN117095152A (en) | 2023-11-21 |
CN117095152B true CN117095152B (en) | 2024-01-26 |
Family
ID=88783582
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311337084.3A Active CN117095152B (en) | 2023-10-17 | 2023-10-17 | Bone recognition camera for physical training evaluation and training evaluation method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117095152B (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110941990A (en) * | 2019-10-22 | 2020-03-31 | 泰康保险集团股份有限公司 | Method and device for evaluating human body actions based on skeleton key points |
CN111680562A (en) * | 2020-05-09 | 2020-09-18 | 北京中广上洋科技股份有限公司 | Human body posture identification method and device based on skeleton key points, storage medium and terminal |
CN112464715A (en) * | 2020-10-22 | 2021-03-09 | 南京理工大学 | Sit-up counting method based on human body bone point detection |
CN113642525A (en) * | 2021-09-02 | 2021-11-12 | 浙江大学 | Infant neural development assessment method and system based on skeletal points |
CN114140530A (en) * | 2021-12-02 | 2022-03-04 | 深圳市火乐科技发展有限公司 | Image processing method and projection equipment |
CN115188074A (en) * | 2022-07-08 | 2022-10-14 | 乐跑体育互联网(武汉)有限公司 | Interactive physical training evaluation method, device and system and computer equipment |
-
2023
- 2023-10-17 CN CN202311337084.3A patent/CN117095152B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110941990A (en) * | 2019-10-22 | 2020-03-31 | 泰康保险集团股份有限公司 | Method and device for evaluating human body actions based on skeleton key points |
CN111680562A (en) * | 2020-05-09 | 2020-09-18 | 北京中广上洋科技股份有限公司 | Human body posture identification method and device based on skeleton key points, storage medium and terminal |
CN112464715A (en) * | 2020-10-22 | 2021-03-09 | 南京理工大学 | Sit-up counting method based on human body bone point detection |
CN113642525A (en) * | 2021-09-02 | 2021-11-12 | 浙江大学 | Infant neural development assessment method and system based on skeletal points |
CN114140530A (en) * | 2021-12-02 | 2022-03-04 | 深圳市火乐科技发展有限公司 | Image processing method and projection equipment |
CN115188074A (en) * | 2022-07-08 | 2022-10-14 | 乐跑体育互联网(武汉)有限公司 | Interactive physical training evaluation method, device and system and computer equipment |
Also Published As
Publication number | Publication date |
---|---|
CN117095152A (en) | 2023-11-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107754225A (en) | A kind of intelligent body-building coaching system | |
CN111368810B (en) | Sit-up detection system and method based on human body and skeleton key point identification | |
US11006856B2 (en) | Method and program product for multi-joint tracking combining embedded sensors and an external sensor | |
CN103099602B (en) | Based on the physical examinations method and system of optical identification | |
CN109692003B (en) | Training system is corrected to children gesture of running | |
CN106923839A (en) | Exercise assist device, exercising support method and recording medium | |
CN111437583A (en) | Badminton basic action auxiliary training system based on Kinect | |
CN105813698A (en) | Fitness training system for merging energy expenditure calculations from multiple devices | |
CN108958482B (en) | Similarity action recognition device and method based on convolutional neural network | |
CN106874874A (en) | Motion state identification method and device | |
CN113384861A (en) | Table tennis training device, table tennis training method, and computer-readable storage medium | |
CN114100103B (en) | Rope skipping counting detection system and method based on key point identification | |
CN113856186B (en) | Pull-up action judging and counting method, system and device | |
CN114255508A (en) | OpenPose-based student posture detection analysis and efficiency evaluation method | |
He et al. | A New Kinect‐Based Posture Recognition Method in Physical Sports Training Based on Urban Data | |
CN117095152B (en) | Bone recognition camera for physical training evaluation and training evaluation method | |
Li et al. | Retracted article: Design and implementation of track and field training information collection and feedback system based on multi-sensor information fusion | |
CN210078765U (en) | Motion capture recognition and evaluation device based on wearable sensor | |
CN114832349B (en) | Yuanzhou swimming teaching auxiliary system and use method thereof | |
CN116805433A (en) | Human motion trail data analysis system | |
CN115050091A (en) | Motion measurement method and motion measurement device based on AI identification | |
CN115809679A (en) | Physical fitness test method, electronic device, storage medium, and computer program product | |
CN113517052A (en) | Multi-perception man-machine interaction system and method in commercial fitness scene | |
CN207804236U (en) | Intelligent physical examination all-in-one machine | |
Jeng | Hierarchical linear model approach to explore interaction effects of swimmers’ characteristics and breathing patterns on swimming performance in butterfly stroke with self-developed inertial measurement unit |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |