CN112464915A - Push-up counting method based on human body bone point detection - Google Patents
Push-up counting method based on human body bone point detection Download PDFInfo
- Publication number
- CN112464915A CN112464915A CN202011606081.1A CN202011606081A CN112464915A CN 112464915 A CN112464915 A CN 112464915A CN 202011606081 A CN202011606081 A CN 202011606081A CN 112464915 A CN112464915 A CN 112464915A
- Authority
- CN
- China
- Prior art keywords
- point
- human body
- convolution
- push
- layers
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 46
- 210000000988 bone and bone Anatomy 0.000 title claims abstract description 32
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000005452 bending Methods 0.000 claims abstract description 5
- 238000005070 sampling Methods 0.000 claims description 30
- 210000003423 ankle Anatomy 0.000 claims description 21
- 210000003127 knee Anatomy 0.000 claims description 20
- 210000000707 wrist Anatomy 0.000 claims description 20
- 230000004927 fusion Effects 0.000 claims description 17
- 238000010586 diagram Methods 0.000 claims description 11
- 238000010606 normalization Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 abstract description 4
- 238000012360 testing method Methods 0.000 description 11
- 238000005516 engineering process Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 210000003205 muscle Anatomy 0.000 description 3
- 230000006870 function Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000003825 pressing Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000010998 test method Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 210000001015 abdomen Anatomy 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 210000000852 deltoid muscle Anatomy 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 210000004247 hand Anatomy 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 210000001364 upper extremity Anatomy 0.000 description 1
Images
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06M—COUNTING MECHANISMS; COUNTING OF OBJECTS NOT OTHERWISE PROVIDED FOR
- G06M1/00—Design features of general application
- G06M1/27—Design features of general application for representing the result of count in the form of electric signals, e.g. by sensing markings on the counter drum
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a push-up counting method based on human body bone point detection, which comprises the steps of obtaining a human body push-up image, marking human body bone point position information, training and obtaining a human body bone point detection network model; shooting the image of the human body of the tested person by a camera, and inputting the obtained image of the human body into the trained human body skeleton point detection network model according to the shooting sequence; acquiring human skeleton point information in the shot picture; judging whether the signal of the detected human skeleton is in a straightened state signal or a straightened state signal according to the position information of the preset skeleton point; if the human body of the tested person has a continuous straightening and bending signal, the person is judged to do push-up counting one. The push-up data acquisition can be realized only by one camera, and the implementation mode is simple and feasible and is easy to popularize; and push-up counting is realized by an image method, so that the manpower condition is reduced.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a push-up counting method based on human body bone point detection.
Background
The push-up is a common body-building exercise, mainly exercises the muscle group of the triceps brachii, the deltoid muscle, the serratus forepawns, the brachial coracoid muscle and other parts of the body, mainly plays a role in improving the muscle strength of the upper limbs, the chest, the waist, the back and the abdomen, and is a simple, easy and effective strength training means. To achieve a perfect starting position for push-ups, the body must be kept straight from the shoulder to the ankle, the arms should be pushed up on the chest with the hands slightly wider than the shoulder, thus ensuring more effective biceps training for each activity.
The prior art application number is CN201610303349, the name of which is: a counting method for push-up test discloses a method for respectively arranging a first identification area and a second identification area with specific lines in a test area; in the test, continuously shooting at a certain frame rate by using a camera arranged obliquely above a test area to obtain a test image comprising a tester, a first texture and a second texture; in each frame of test image from the initial frame to the later frame, judging whether the minimum value of the pixel value of the first texture is smaller than the low threshold value in the process of changing from reduction to increase, and then judging whether the maximum value of the pixel value of the first texture is larger than the high threshold value in the process of changing from increase to decrease, if so, judging whether the pixel values of the second texture are larger than the kneeling posture threshold value in the judging period, and if so, counting once. The error rate of the pixel value comparison method adopted by the method is higher, and the method has higher requirements on the acquisition angle of the camera.
The invention name of prior art application number CN202010081035.8 is: a push-up test method and a device thereof disclose a rule judgment and performance measurement for push-up test by using a camera and a machine vision technology, which breaks through the limitation that the traditional push-up test technology can not carry out intelligent test, and the intellectualization of the push-up test is realized by introducing the camera and the machine vision technology into the push-up test method and the device, thereby improving the efficiency of the push-up test and the accuracy of the measurement of the push-up performance. But the count is also more demanding on the acquisition of pictures.
The traditional push-up counting device mainly has a push-up counting mode and two push-up counting modes based on wearable equipment. Based on push type counting mode, judge counting mode through pressing, need be complicated at the site operation, can not open-air erection equipment, receive sleet weather influence great, probably lead to pressing equipment or sensor to damage. According to the control method and device based on the wearable equipment, a user needs to wear the device, so that the user experience is poor.
Disclosure of Invention
1. The technical problem to be solved is as follows:
aiming at the technical problem, the invention provides a push-up counting method based on human body bone point detection.
2. The technical scheme is as follows:
a push-up identification method based on human body identification is characterized in that: the method comprises the following steps:
the method comprises the following steps: acquiring a picture containing a human body push-up image, extracting and marking human body bone point location information from human body push-up image data so as to obtain a human body bone point location data set; inputting the marked push-up picture and the marked truth value data into a human body bone point detection network to train and obtain a human body bone point detection network model, wherein the human body bone point detection network is a human body bone point detection network based on multi-scale fusion.
Step two: shooting the image of the human body of the tested person by a camera, and inputting the obtained image of the human body into the trained human body skeleton point detection network model according to the shooting sequence; acquiring human skeleton point information in the shot picture;
step three: judging a signal of the detected human skeleton according to the position information of the preset skeleton point; the obtained human skeleton point information comprises positions of a right shoulder point, a right elbow point, a right wrist point, a left shoulder point, a left elbow point, a left wrist point, a right hip point, a right knee point, a right ankle point, a left hip point, a left knee point and a left ankle point; the preset skeleton point position information is that the break angle of a broken line formed by connecting a right shoulder point, a right elbow point and a right wrist point of a measured person is more than 150 degrees, the break angle of a broken line formed by connecting a left shoulder point, a left elbow point and a left wrist point is more than 150 degrees, the break angle of a broken line formed by connecting a right hip point, a right knee point and a right ankle point is more than 170 degrees, and the break angle of a broken line formed by connecting a left hip point, a left knee point and a left ankle point is more than 170 degrees, so that a signal that the human body is in a straight state is judged; presetting that the break angle of a broken line formed by connecting a right shoulder point, a right elbow point and a right wrist point of a tested person is less than 30 degrees, the break angle of a broken line formed by connecting a left shoulder point, a left elbow point and a left wrist point is less than 30 degrees, the break angle of a broken line formed by connecting a right hip point, a right knee point and a right ankle point is more than 170 degrees, and the break angle of a broken line formed by connecting a left hip point, a left knee point and a left ankle point is more than 170 degrees, and then judging that the human body is in a straightening state signal.
Step four: if the human body of the tested person has a continuous straightening and bending signal, the person is judged to do push-up counting one.
Further, the step one of acquiring the human body push-up image data is specifically to acquire a video and a picture for making push-up in a network; cutting each frame of the obtained video into an image; and mixing the cut image and the image obtained by the original crawler, marking out the person in the image, cutting out the marked person, uniformly performing normalization processing, and then performing skeleton point location marking.
Further, the human body skeleton point detection network comprises a backbone network, a feature fusion network and an output network.
Further, a backbone network of the human skeleton point detection network uses resnet101 as the backbone network, wherein the resnet101 is composed of 5 convolution modules; the first convolution module of resnet101 consists of three repeated convolution groups, each convolution group containing 64 1 × 1 convolution layers, 64 3 × 3 convolution layers and 256 1 × 1 convolution layers; the second convolution module of resnet101 consists of four repeated convolution groups, each convolution group containing 128 1 × 1 convolution layers, 128 3 × 3 convolution layers and 512 1 × 1 convolution layers; the third convolution module of resnet101 consists of twenty-three repeating convolution groups, each convolution group containing 256 1 x 1 convolutional layers, 256 3 x 3 convolutional layers, and 1024 1 x 1 convolutional layers; the fourth convolution module of resnet101 consists of three repeated convolution groups, each convolution group containing 512 1 × 1 convolution layers, 512 3 × 3 convolution layers and 2048 1 × 1 convolution layers.
Furthermore, the feature fusion network of the human body bone point detection network is a high-low layer fusion structure realized by fusing three times of down-sampling, four times of down-sampling and five times of down-sampling scale feature maps; for triple down-sampling, 3 × 3 convolution operation with the step length of 2 is firstly carried out on a triple down-sampling layer to realize down-sampling, then 1 × 1 convolution layer is used to realize channel multiplication, the number and the dimensionality of the channels are the same as those of a quadruple down-sampling layer, for a quintuple down-sampling layer, the number of the channels is reduced to half of the original number through 1 × 1 convolution, and then the size multiplication of a characteristic layer is realized through up-sampling; connecting the feature layers obtained by the three scales, and finally performing 3-by-3 convolution on the connected feature layers to generate a feature layer A; and respectively performing 1 × 1 convolution on the feature layer A, respectively performing 3 × 3 convolution and 1 × 1 convolution to generate a feature layer B, respectively performing 1 × 1 convolution and 2 fully-connected layers on the feature layer A to generate a feature layer C, and multiplying A and C to obtain a heat point diagram D.
Further, the output layer of the human body skeleton point detection network outputs key point coordinates and coordinates classification through a hotspot graph D to output key points of the key points, and the prediction loss function uses MSE loss classification and uses mutual entropy loss.
3. Has the advantages that:
(1) the push-up data acquisition of the testee can be realized only by one camera, and the implementation mode is simple and feasible and is easy to popularize.
(2) The push-up counting is realized by an image method, so that the manpower condition is reduced.
(3) According to the invention, the multi-scale feature layers are fused and then feature selection is carried out, so that the network output layer information not only has deep semantic information, but also has low-level position information.
(4) The invention sends the attention mechanism to the multi-scale fusion layer, so that the position of the human skeleton point which can be learned by the network is more accurate.
(5) In the invention, the human skeleton point detection network model is trained by using the multilayer convolutional neural network model in the deep learning model, so that the detection is more accurate. .
Drawings
FIG. 1 is a schematic diagram of the present invention for collecting the picture of the tested person;
FIG. 2 is a diagram of human bone points in accordance with the present invention;
FIG. 3 is a flow chart of a detection network model for training human skeletal points in the present invention;
FIG. 4 is a schematic view of a human skeletal point detection network attention mechanism according to the present invention; (ii) a
FIG. 5 is a schematic structural diagram of a feature fusion network of the human skeletal point detection network according to the present invention;
FIG. 6 is a flow chart of the present invention.
Detailed Description
The present invention will be described in detail with reference to the accompanying drawings.
As shown in the attached drawings 1 to 5, a push-up recognition method based on human body recognition is characterized in that: the method comprises the following steps:
the method comprises the following steps: acquiring a picture containing a human body push-up image, extracting and marking human body bone point location information from human body push-up image data so as to obtain a human body bone point location data set; inputting the marked push-up picture and the marked truth value data into a human body bone point detection network to train and obtain a human body bone point detection network model, wherein the human body bone point detection network is a human body bone point detection network based on multi-scale fusion.
Step two: shooting the image of the human body of the tested person by a camera, and inputting the obtained image of the human body into the trained human body skeleton point detection network model according to the shooting sequence; acquiring human skeleton point information in the shot picture;
step three: judging a signal of the detected human skeleton according to the position information of the preset skeleton point; the obtained human skeleton point information comprises positions of a right shoulder point, a right elbow point, a right wrist point, a left shoulder point, a left elbow point, a left wrist point, a right hip point, a right knee point, a right ankle point, a left hip point, a left knee point and a left ankle point; the preset skeleton point position information is that the break angle of a broken line formed by connecting a right shoulder point, a right elbow point and a right wrist point of a measured person is more than 150 degrees, the break angle of a broken line formed by connecting a left shoulder point, a left elbow point and a left wrist point is more than 150 degrees, the break angle of a broken line formed by connecting a right hip point, a right knee point and a right ankle point is more than 170 degrees, and the break angle of a broken line formed by connecting a left hip point, a left knee point and a left ankle point is more than 170 degrees, so that a signal that the human body is in a straight state is judged; presetting that the break angle of a broken line formed by connecting a right shoulder point, a right elbow point and a right wrist point of a tested person is less than 30 degrees, the break angle of a broken line formed by connecting a left shoulder point, a left elbow point and a left wrist point is less than 30 degrees, the break angle of a broken line formed by connecting a right hip point, a right knee point and a right ankle point is more than 170 degrees, and the break angle of a broken line formed by connecting a left hip point, a left knee point and a left ankle point is more than 170 degrees, and then judging that the human body is in a straightening state signal.
Step four: if the human body of the tested person has a continuous straightening and bending signal, the person is judged to do push-up counting one.
Further, the step one of acquiring the human body push-up image data is specifically to acquire a video and a picture for making push-up in a network; cutting each frame of the obtained video into an image; and mixing the cut image and the image obtained by the original crawler, marking out the person in the image, cutting out the marked person, uniformly performing normalization processing, and then performing skeleton point location marking.
Further, the human body skeleton point detection network comprises a backbone network, a feature fusion network and an output network.
Further, a backbone network of the human skeleton point detection network uses resnet101 as the backbone network, wherein the resnet101 is composed of 5 convolution modules; the first convolution module of resnet101 consists of three repeated convolution groups, each convolution group containing 64 1 × 1 convolution layers, 64 3 × 3 convolution layers and 256 1 × 1 convolution layers; the second convolution module of resnet101 consists of four repeated convolution groups, each convolution group containing 128 1 × 1 convolution layers, 128 3 × 3 convolution layers and 512 1 × 1 convolution layers; the third convolution module of resnet101 consists of twenty-three repeating convolution groups, each convolution group containing 256 1 x 1 convolutional layers, 256 3 x 3 convolutional layers, and 1024 1 x 1 convolutional layers; the fourth convolution module of resnet101 consists of three repeated convolution groups, each convolution group containing 512 1 × 1 convolution layers, 512 3 × 3 convolution layers and 2048 1 × 1 convolution layers.
Furthermore, the feature fusion network of the human body bone point detection network is a high-low layer fusion structure realized by fusing three times of down-sampling, four times of down-sampling and five times of down-sampling scale feature maps; for triple down-sampling, 3 × 3 convolution operation with the step length of 2 is firstly carried out on a triple down-sampling layer to realize down-sampling, then 1 × 1 convolution layer is used to realize channel multiplication, the number and the dimensionality of the channels are the same as those of a quadruple down-sampling layer, for a quintuple down-sampling layer, the number of the channels is reduced to half of the original number through 1 × 1 convolution, and then the size multiplication of a characteristic layer is realized through up-sampling; connecting the feature layers obtained by the three scales, and finally performing 3-by-3 convolution on the connected feature layers to generate a feature layer A; and respectively performing 1 × 1 convolution on the feature layer A, respectively performing 3 × 3 convolution and 1 × 1 convolution to generate a feature layer B, respectively performing 1 × 1 convolution and 2 fully-connected layers on the feature layer A to generate a feature layer C, and multiplying A and C to obtain a heat point diagram D.
Further, the output layer of the human body skeleton point detection network outputs key point coordinates and coordinates classification through a hotspot graph D to output key points of the key points, and the prediction loss function uses MSE loss classification and uses mutual entropy loss.
The specific embodiment is as follows:
as shown in fig. 1, the image acquisition of the subject needs an imaging unit capable of realizing the position of the subject, and fig. 1 is an optional implementation manner in a schematic diagram thereof, in which a camera is installed obliquely downward, and an image set of the subject can be acquired by the camera.
As shown in fig. 2, in the present invention, a human skeleton point detection network model needs to be established first, and when a picture including a human body is input, human skeleton points in the picture can be output quickly and accurately. The skeleton point bitmap of the human body in the invention records 18 point positions, wherein, 0 represents nose, 1 represents neck, 2 represents right shoulder, 3 represents right elbow, 4 represents right wrist, 5 represents left shoulder, 6 represents left elbow, 7 represents left wrist, 8 represents right hip, 9 represents right knee, 10 represents right ankle, 11 represents left hip, 12 represents left knee, 13 left ankle, 14 represents right eye, 15 represents left eye, 16 represents right ear, and 17 represents left ear.
As shown in fig. 3, the human bone point detection network model of the invention obtains a picture containing a human push-up image, extracts and marks human point location information from the human push-up image data to obtain a human bone point data set; and inputting the marked push-up picture and the marked truth value data into a human body bone point detection network to train and obtain a human body bone point detection network model. When a person prepares for push-up, when the point 2, the point 3 and the point 4 are sequentially connected to form a break angle, the break angle is larger than 150 degrees, the point 5, the point 6 and the point 7 are sequentially connected to form a break angle, the break angle is larger than 150 degrees, the point 8, the point 9 and the point 10 are sequentially connected to form a break angle, the break angle is larger than 170 degrees, the point 11, the point 12 and the point 13 are sequentially connected to form a break angle, the break angle is larger than 170 degrees, a signal in a straightening state is sent out, when the point 2, the point 3 and the point 4 are sequentially connected to form a break angle, the break angle is smaller than 30 degrees, the point 5, the point 6 and the point 7 are smaller than 30 degrees, the point 8, the point 9 and the point 10 are sequentially connected to form a break angle, the break angle is larger than 170 degrees, the point 11, the point 12 and the point 13 are sequentially connected to form a break angle, the break angle is larger. When a continuous straightening and bending signal is identified, the count of the push-ups is increased by 1.
Fig. 4 is a schematic structural diagram of a human skeleton point detection network attention mechanism, namely, a structural diagram of resnet 101; as shown, where resnet101 consists of 5 convolution modules; the first convolution module of resnet101 consists of three repeated convolution groups, each convolution group containing 64 1 × 1 convolution layers, 64 3 × 3 convolution layers and 256 1 × 1 convolution layers; the second convolution module of resnet101 consists of four repeated convolution groups, each convolution group containing 128 1 × 1 convolution layers, 128 3 × 3 convolution layers and 512 1 × 1 convolution layers; the third convolution module of resnet101 consists of twenty-three repeating convolution groups, each convolution group containing 256 1 x 1 convolutional layers, 256 3 x 3 convolutional layers, and 1024 1 x 1 convolutional layers; the fourth convolution module of resnet101 consists of three repeated convolution groups, each convolution group containing 512 1 × 1 convolution layers, 512 3 × 3 convolution layers and 2048 1 × 1 convolution layers.
FIG. 5 is a schematic structural diagram of a feature fusion network of the human skeletal point detection network according to the present invention; the feature fusion network of the human body skeleton point detection network is a high-low layer fusion structure realized by fusing three times of downsampling, four times of downsampling and five times of downsampling scale feature maps; for triple down-sampling, 3 × 3 convolution operation with the step length of 2 is firstly carried out on a triple down-sampling layer to realize down-sampling, then 1 × 1 convolution layer is used to realize channel multiplication, the number and the dimensionality of the channels are the same as those of a quadruple down-sampling layer, for a quintuple down-sampling layer, the number of the channels is reduced to half of the original number through 1 × 1 convolution, and then the size multiplication of a characteristic layer is realized through up-sampling; connecting the feature layers obtained by the three scales, and finally performing 3-by-3 convolution on the connected feature layers to generate a feature layer A; and respectively performing 1 × 1 convolution on the feature layer A, respectively performing 3 × 3 convolution and 1 × 1 convolution to generate a feature layer B, respectively performing 1 × 1 convolution and 2 fully-connected layers on the feature layer A to generate a feature layer C, and multiplying A and C to obtain a heat point diagram D.
Although the present invention has been described with reference to the preferred embodiments, it should be understood that various changes and modifications can be made therein by those skilled in the art without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (6)
1. A push-up identification method based on human body identification is characterized in that: the method comprises the following steps:
the method comprises the following steps: acquiring a picture containing a human body push-up image, extracting and marking human body bone point location information from human body push-up image data so as to obtain a human body bone point location data set; inputting the marked push-up picture and the marked truth value data into a human body bone point detection network to train and obtain a human body bone point detection network model, wherein the human body bone point detection network is a human body bone point detection network based on multi-scale fusion;
step two: shooting the image of the human body of the tested person by a camera, and inputting the obtained image of the human body into the trained human body skeleton point detection network model according to the shooting sequence; acquiring human skeleton point information in the shot picture;
step three: judging a signal of the detected human skeleton according to the position information of the preset skeleton point; the obtained human skeleton point information comprises positions of a right shoulder point, a right elbow point, a right wrist point, a left shoulder point, a left elbow point, a left wrist point, a right hip point, a right knee point, a right ankle point, a left hip point, a left knee point and a left ankle point; the preset skeleton point position information is that the break angle of a broken line formed by connecting a right shoulder point, a right elbow point and a right wrist point of a measured person is more than 150 degrees, the break angle of a broken line formed by connecting a left shoulder point, a left elbow point and a left wrist point is more than 150 degrees, the break angle of a broken line formed by connecting a right hip point, a right knee point and a right ankle point is more than 170 degrees, and the break angle of a broken line formed by connecting a left hip point, a left knee point and a left ankle point is more than 170 degrees, so that a signal that the human body is in a straight state is judged; presetting a folding angle of a folding line formed by connecting a right shoulder point, a right elbow point and a right wrist point of a tested person to be less than 30 degrees, a folding angle of a folding line formed by connecting a left shoulder point, a left elbow point and a left wrist point to be less than 30 degrees, a folding angle of a folding line formed by connecting a right hip point, a right knee point and a right ankle point to be more than 170 degrees, and a folding angle of a folding line formed by connecting a left hip point, a left knee point and a left ankle point to be more than 170 degrees, and judging that the human body is in a straightening state signal;
step four: if the human body of the tested person has a continuous straightening and bending signal, the person is judged to do push-up counting one.
2. The push-up recognition method based on human body recognition according to claim 1, characterized in that: acquiring human body push-up image data in the first step, specifically acquiring videos and pictures for making push-up in a network; cutting each frame of the obtained video into an image; and mixing the cut image and the image obtained by the original crawler, marking out the person in the image, cutting out the marked person, uniformly performing normalization processing, and then performing skeleton point location marking.
3. The push-up recognition method based on human body recognition according to claim 1, characterized in that: the human body skeleton point detection network comprises a backbone network, a feature fusion network and an output network.
4. The push-up recognition method based on human body recognition according to claim 3, characterized in that: the backbone network of the human skeleton point detection network uses a resnet101 as the backbone network, wherein the resnet101 is composed of 5 convolution modules; the first convolution module of resnet101 consists of three repeated convolution groups, each convolution group containing 64 1 × 1 convolution layers, 64 3 × 3 convolution layers and 256 1 × 1 convolution layers; the second convolution module of resnet101 consists of four repeated convolution groups, each convolution group containing 128 1 × 1 convolution layers, 128 3 × 3 convolution layers and 512 1 × 1 convolution layers; the third convolution module of resnet101 consists of twenty-three repeating convolution groups, each convolution group containing 256 1 x 1 convolutional layers, 256 3 x 3 convolutional layers, and 1024 1 x 1 convolutional layers; the fourth convolution module of resnet101 consists of three repeated convolution groups, each convolution group containing 512 1 × 1 convolution layers, 512 3 × 3 convolution layers and 2048 1 × 1 convolution layers.
5. The push-up recognition method based on human body recognition according to claim 3, characterized in that: the feature fusion network of the human body skeleton point detection network is a high-low layer fusion structure realized by fusing three times of downsampling, four times of downsampling and five times of downsampling scale feature maps; for triple down-sampling, 3 × 3 convolution operation with the step length of 2 is firstly carried out on a triple down-sampling layer to realize down-sampling, then 1 × 1 convolution layer is used to realize channel multiplication, the number and the dimensionality of the channels are the same as those of a quadruple down-sampling layer, for a quintuple down-sampling layer, the number of the channels is reduced to half of the original number through 1 × 1 convolution, and then the size multiplication of a characteristic layer is realized through up-sampling; connecting the feature layers obtained by the three scales, and finally performing 3-by-3 convolution on the connected feature layers to generate a feature layer A; and respectively performing 1 × 1 convolution on the feature layer A, respectively performing 3 × 3 convolution and 1 × 1 convolution to generate a feature layer B, respectively performing 1 × 1 convolution and 2 fully-connected layers on the feature layer A to generate a feature layer C, and multiplying A and C to obtain a heat point diagram D.
6. The push-up recognition method based on human body recognition according to claim 5, characterized in that: the output layer of the human body skeleton point detection network outputs the coordinates of the key points through the hot spot graph D and realizes the output of the key points through coordinate classification, and the prediction loss function uses the MSE loss classification and uses the mutual entropy loss.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011606081.1A CN112464915B (en) | 2020-12-30 | 2020-12-30 | Push-up counting method based on human skeleton point detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011606081.1A CN112464915B (en) | 2020-12-30 | 2020-12-30 | Push-up counting method based on human skeleton point detection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112464915A true CN112464915A (en) | 2021-03-09 |
| CN112464915B CN112464915B (en) | 2024-03-26 |
Family
ID=74802207
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011606081.1A Active CN112464915B (en) | 2020-12-30 | 2020-12-30 | Push-up counting method based on human skeleton point detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112464915B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113191200A (en) * | 2021-04-06 | 2021-07-30 | 恒鸿达科技有限公司 | Push-up test counting method, device, equipment and medium |
| CN113255623A (en) * | 2021-07-14 | 2021-08-13 | 北京壹体科技有限公司 | System and method for intelligently identifying push-up action posture completion condition |
| CN113599776A (en) * | 2021-08-05 | 2021-11-05 | 北京理工大学 | Real-time push-up counting and standard judging method and system |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101172200A (en) * | 2006-07-18 | 2008-05-07 | 孙学川 | Intelligent push-up test system |
| CN101468256A (en) * | 2008-08-13 | 2009-07-01 | 孙学川 | Push-up intelligent tester |
| CN101934135A (en) * | 2010-07-26 | 2011-01-05 | 南京理工大学 | Physical ability automated test system |
| CN106934830A (en) * | 2017-03-14 | 2017-07-07 | 北京林业大学 | A kind of contactless fitness test system and method for testing based on depth image |
| CN206463477U (en) * | 2017-01-18 | 2017-09-05 | 中国人民解放军装甲兵工程学院 | Push-up test system |
| CN206965078U (en) * | 2017-05-24 | 2018-02-06 | 中国人民武装警察部队警官学院 | Military anti-cheating standard push-up examination device |
| CN108379819A (en) * | 2018-04-28 | 2018-08-10 | 广西职业技术学院 | A kind of push-up counting device |
| CN110298221A (en) * | 2018-03-23 | 2019-10-01 | 上海形趣信息科技有限公司 | Self-service body building method, system, electronic equipment, storage medium |
| CN111275023A (en) * | 2020-03-19 | 2020-06-12 | 中国人民解放军国防科技大学 | Push-up test system based on face recognition and human body posture estimation |
| CN111368810A (en) * | 2020-05-26 | 2020-07-03 | 西南交通大学 | Sit-up detection system and method based on human body and skeleton key point identification |
| CN111428664A (en) * | 2020-03-30 | 2020-07-17 | 厦门瑞为信息技术有限公司 | Real-time multi-person posture estimation method based on artificial intelligence deep learning technology for computer vision |
| CN111597975A (en) * | 2020-05-14 | 2020-08-28 | 北京万觉科技有限公司 | Personnel action detection method and device and electronic equipment |
| CN111967305A (en) * | 2020-07-01 | 2020-11-20 | 华南理工大学 | Real-time multi-scale target detection method based on lightweight convolutional neural network |
-
2020
- 2020-12-30 CN CN202011606081.1A patent/CN112464915B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101172200A (en) * | 2006-07-18 | 2008-05-07 | 孙学川 | Intelligent push-up test system |
| CN101468256A (en) * | 2008-08-13 | 2009-07-01 | 孙学川 | Push-up intelligent tester |
| CN101934135A (en) * | 2010-07-26 | 2011-01-05 | 南京理工大学 | Physical ability automated test system |
| CN206463477U (en) * | 2017-01-18 | 2017-09-05 | 中国人民解放军装甲兵工程学院 | Push-up test system |
| CN106934830A (en) * | 2017-03-14 | 2017-07-07 | 北京林业大学 | A kind of contactless fitness test system and method for testing based on depth image |
| CN206965078U (en) * | 2017-05-24 | 2018-02-06 | 中国人民武装警察部队警官学院 | Military anti-cheating standard push-up examination device |
| CN110298221A (en) * | 2018-03-23 | 2019-10-01 | 上海形趣信息科技有限公司 | Self-service body building method, system, electronic equipment, storage medium |
| CN108379819A (en) * | 2018-04-28 | 2018-08-10 | 广西职业技术学院 | A kind of push-up counting device |
| CN111275023A (en) * | 2020-03-19 | 2020-06-12 | 中国人民解放军国防科技大学 | Push-up test system based on face recognition and human body posture estimation |
| CN111428664A (en) * | 2020-03-30 | 2020-07-17 | 厦门瑞为信息技术有限公司 | Real-time multi-person posture estimation method based on artificial intelligence deep learning technology for computer vision |
| CN111597975A (en) * | 2020-05-14 | 2020-08-28 | 北京万觉科技有限公司 | Personnel action detection method and device and electronic equipment |
| CN111368810A (en) * | 2020-05-26 | 2020-07-03 | 西南交通大学 | Sit-up detection system and method based on human body and skeleton key point identification |
| CN111967305A (en) * | 2020-07-01 | 2020-11-20 | 华南理工大学 | Real-time multi-scale target detection method based on lightweight convolutional neural network |
Non-Patent Citations (1)
| Title |
|---|
| 李岳: "基于深度学习的人体姿态识别方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 08 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113191200A (en) * | 2021-04-06 | 2021-07-30 | 恒鸿达科技有限公司 | Push-up test counting method, device, equipment and medium |
| CN113255623A (en) * | 2021-07-14 | 2021-08-13 | 北京壹体科技有限公司 | System and method for intelligently identifying push-up action posture completion condition |
| CN113599776A (en) * | 2021-08-05 | 2021-11-05 | 北京理工大学 | Real-time push-up counting and standard judging method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112464915B (en) | 2024-03-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112464915B (en) | Push-up counting method based on human skeleton point detection | |
| CN108734104B (en) | Body-building action error correction method and system based on deep learning image recognition | |
| CN109934111B (en) | Fitness posture estimation method and system based on key points | |
| WO2021057810A1 (en) | Data processing method, data training method, data identifying method and device, and storage medium | |
| CN111368810B (en) | Sit-up detection system and method based on human body and skeleton key point identification | |
| CN109815907B (en) | Sit-up posture detection and guidance method based on computer vision technology | |
| CN109410261B (en) | Monocular image depth estimation method based on pyramid pooling module | |
| CN111563887B (en) | Intelligent analysis method and device for oral cavity image | |
| CN110728220A (en) | Gymnastics auxiliary training method based on human body action skeleton information | |
| CN109753891A (en) | Football player's orientation calibration method and system based on human body critical point detection | |
| CN109872305A (en) | It is a kind of based on Quality Map generate network without reference stereo image quality evaluation method | |
| CN110246148A (en) | The conspicuousness detection method of multi-modal depth information fusion and attention study | |
| CN112364785A (en) | Exercise training guidance method, device, equipment and computer storage medium | |
| CN111401260A (en) | Sit-up test counting method and system based on Quick-OpenPose model | |
| CN111833439B (en) | Artificial intelligence based ammunition throwing analysis and mobile simulation training method | |
| CN111466878A (en) | Real-time monitoring method and device for pain symptoms of bedridden patients based on expression recognition | |
| CN111912531A (en) | Dynamic human body trunk temperature analysis method based on intelligent thermal imaging video analysis | |
| CN115953834A (en) | Multi-head attention posture estimation method and detection system for sit-up | |
| CN114998803A (en) | Body-building movement classification and counting method based on video | |
| CN113657266B (en) | Fitness training management method and system based on intelligent bracelet and human body three-dimensional reconstruction | |
| CN115690895A (en) | Human skeleton point detection-based multi-person motion detection method and device | |
| CN111031577B (en) | Multi-node wireless motion capture node expansion method | |
| CN112784699B (en) | Implementation method and system for assessment and guidance of exercise training gestures | |
| CN105611285B (en) | General non-reference picture quality appraisement method based on phase selective mechanism | |
| Liang et al. | Research on Fitness Action Evaluation System Based on Skeleton |
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 |