CN117031434B - Real-time falling detection method based on millimeter wave radar - Google Patents
Real-time falling detection method based on millimeter wave radar Download PDFInfo
- Publication number
- CN117031434B CN117031434B CN202311291588.6A CN202311291588A CN117031434B CN 117031434 B CN117031434 B CN 117031434B CN 202311291588 A CN202311291588 A CN 202311291588A CN 117031434 B CN117031434 B CN 117031434B
- Authority
- CN
- China
- Prior art keywords
- angle
- millimeter wave
- representing
- wave radar
- distance
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 93
- 238000012549 training Methods 0.000 claims abstract description 107
- 238000000034 method Methods 0.000 claims abstract description 39
- 238000012545 processing Methods 0.000 claims abstract description 23
- 239000011159 matrix material Substances 0.000 claims abstract description 16
- 230000006870 function Effects 0.000 claims description 30
- 230000001629 suppression Effects 0.000 claims description 13
- 238000013528 artificial neural network Methods 0.000 claims description 12
- 230000007246 mechanism Effects 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 8
- 230000005540 biological transmission Effects 0.000 claims description 7
- 238000003672 processing method Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 5
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 238000013519 translation Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 15
- 230000015654 memory Effects 0.000 description 9
- 230000007613 environmental effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 230000033228 biological regulation Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000011897 real-time detection Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000013475 authorization Methods 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
- G01S7/417—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section involving the use of neural networks
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/886—Radar or analogous systems specially adapted for specific applications for alarm systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computer Networks & Wireless Communication (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Psychiatry (AREA)
- Electromagnetism (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention provides a real-time fall detection method based on millimeter wave radar. The method comprises the following steps: mixing the transmitting signal and the receiving signal of the millimeter wave radar to obtain an intermediate frequency signal; performing fast Fourier transform on the intermediate frequency signals along the distance dimension, the speed dimension and the angle dimension respectively to obtain distance information, speed information and angle information, and obtaining a distance-Doppler-angle three-dimensional matrix based on the distance information, the speed information and the angle information; based on a preset weighted summation formula, carrying out weighted summation on the distance-Doppler-angle three-dimensional matrix in the speed dimension to obtain a weighted distance angle heat map, and carrying out image superposition on the weighted distance angle heat map of a plurality of continuous frames in the channel dimension to obtain an image superposition result; dividing the image superposition result according to a preset dividing rule, and processing the image division result by using a real-time fall detection model after training to obtain a fall detection result of the target to be detected.
Description
Technical Field
The invention relates to the field of wireless human body intelligent perception, in particular to a real-time falling detection method based on millimeter wave radar, a training method of a real-time falling detection model, electronic equipment and a storage medium.
Background
Medical care problems for the elderly and other specific populations are becoming more and more of a concern, with accidental falls being one of the major threats to the safety of the elderly. To solve this problem, wearable devices based on speed recognition are often used to detect the posture of the elderly, providing help in time. The devices can monitor the movement and the gesture of the elderly in real time, and analyze and identify whether a falling event occurs through an algorithm. Upon detection of a fall, the device may trigger an alarm or send a notification so that action can be taken quickly and assistance provided.
However, in the prior art, the fall detection method based on speed recognition is easy to generate false alarm when performing fall detection, and other technical schemes, such as the fall detection method based on wireless signals, have higher dependence on different environments and radar deployment methods.
Disclosure of Invention
In view of the above, the present invention provides a real-time fall detection method based on millimeter wave radar, a training method of a real-time fall detection model, an electronic device, and a storage medium, in order to solve at least one of the above problems.
According to a first aspect of the present invention, there is provided a method of real-time fall detection based on millimeter wave radar, comprising:
mixing the transmitting signal and the receiving signal of the millimeter wave radar to obtain an intermediate frequency signal, wherein the receiving signal is a radar echo signal reflected by a target to be detected;
performing fast Fourier transform on the intermediate frequency signals along the distance dimension, the speed dimension and the angle dimension respectively to obtain distance information, speed information and angle information respectively, and obtaining a distance-Doppler-angle three-dimensional matrix based on the distance information, the speed information and the angle information;
based on a preset weighted summation formula, carrying out weighted summation on the distance-Doppler-angle three-dimensional matrix in the speed dimension to obtain a weighted distance-angle heat map, and carrying out image superposition on the weighted distance-angle heat map of continuous multiframes in the channel dimension to obtain an image superposition result;
dividing the image superposition result according to a preset dividing rule, and processing the image division result by using a real-time fall detection model which is trained according to a false alarm suppression mechanism to obtain a fall detection result of the target to be detected.
According to an embodiment of the present invention, the above distance information is represented by formula (1):
（1），
wherein the speed information is represented by formula (2):
（2），
wherein the angle information is represented by formula (3):
（3），
wherein the preset weighted sum formula is represented by formula (4):
（4），
wherein,characterizing the frequency of the intermediate frequency signal, ">Characterizing the slope of the intermediate frequency signal->Characterizing the speed of light->Characterizing the wavelength of the intermediate frequency signal, ">Characterizing the phase difference between two adjacent intermediate frequency signals, a ∈>Characterizing the time difference between two adjacent intermediate frequency signals, a ∈>Characterizing the distance between two adjacent receive antennas of a millimeter wave radar,/->Representing the power of the intermediate frequency signal after fast fourier transformation,/and the like>And->Respectively representing different weight super parameters related to the speed information.
According to the embodiment of the invention, the false alarm suppression mechanism indicates that the object to be detected falls when a plurality of falls are detected in a plurality of continuous time windows, and the object to be detected falls when a plurality of falls are not detected in a plurality of continuous time windows, so that the object to be detected falls into false alarm;
wherein the probability of false alarm is represented by formula (5):
（5），
wherein,classification result output by the real-time fall detection model representing training completion,/->Indicating the time-of-day position of the time window in which the object to be detected falls.
According to a second aspect of the present invention, there is provided a training method of a real-time fall detection model, applied to a real-time fall detection method based on millimeter wave radar, comprising:
preprocessing a training transmitting signal and a training receiving signal of the millimeter wave radar to obtain a training sample image, and carrying out data enhancement on the training sample image to obtain an enhanced training sample image, wherein the training receiving signal is a radar echo signal reflected by a target to be detected;
processing the enhanced training sample image by a signal processing method to construct a positive sample pair with a truth value label, and processing the positive sample pair by utilizing a real-time fall detection model to obtain a detection result, wherein the real-time fall detection model is constructed based on a contrast learning neural network and a classification neural network;
processing the detection result and the truth value label of the positive sample pair by using a preset loss function to obtain a loss value, and carrying out parameter optimization and updating on the real-time falling detection model according to the loss value;
and carrying out data preprocessing, data enhancement, positive sample pair construction, model processing, loss calculation and parameter optimization and updating operation iteratively until preset training conditions are met, so as to obtain a real-time fall detection model after training is completed.
According to an embodiment of the present invention, preprocessing the training transmission signal and the training reception signal of the millimeter wave radar to obtain a training sample image includes:
mixing a training transmitting signal and a training receiving signal of the millimeter wave radar to obtain a training intermediate frequency signal, and respectively performing fast Fourier transformation on the training intermediate frequency signal along a speed dimension, a distance dimension and an angle dimension to obtain speed information, distance information and angle information;
and obtaining a distance-Doppler-angle three-dimensional matrix based on the speed information, the distance information and the angle information, carrying out weighted summation on the distance-Doppler-angle three-dimensional matrix in the speed dimension, and carrying out image superposition on the obtained weighted distance-angle heat map of the continuous multiframes in the channel dimension to obtain a training sample image.
According to an embodiment of the present invention, the data enhancing the training sample image to obtain an enhanced training sample image includes:
carrying out data enhancement on the training sample image through image inversion operation, image translation operation and image frame extraction operation to obtain a data enhanced training sample image;
and carrying out data enhancement on the angle dimension on the training sample image with the data enhancement through a re-weighting operation, so as to obtain an enhanced training sample image.
According to the embodiment of the invention, the data enhancement is performed on the data-enhanced training sample image in the angle dimension through the re-weighting operation, so that the enhanced training sample image comprises angle resolution data enhancement, signal strength data enhancement, data enhancement of vertical tangential angle offset and horizontal tangential angle offset;
wherein the angular resolution data enhancement is represented by formula (6):
（6），
wherein the signal strength data enhancement is represented by formula (7):
（7），
wherein the data enhancement of the vertical tangential angle offset and the horizontal tangential angle offset is represented by formula (8):
（8），
wherein,characterizing the wavelength of the intermediate frequency signal, ">Representing the distance between adjacent antennas of a millimeter wave radar, +.>Representing the included angle between the object to be detected and the right direction of the millimeter wave radar, and the angle is +>Representing the signal power of the millimeter wave radar transmit antenna,represents the gain of the millimeter wave radar transmitting antenna, +.>Represents the gain of the millimeter wave radar receiving antenna, +.>Representing the wavelength of the millimeter wave radar transmission signal, +.>Representing the number of antennas of a millimeter wave radar, +.>And->Representing coordinates of the millimeter wave radar in a three-dimensional space; />Representing the scattering cross section of a millimeter wave radar by adjusting +.>Can achieve data enhancement in the angular dimension, +.>Expressed in (/ -)>) Center of rotation pair->Angle of rotation, (-)>) Representing the original spatial coordinates of the object to be detected, +.>Representing the spatial coordinates of the object to be detected after rotation, < > and>representing the distance between the object to be detected and the millimeter wave radar.
According to an embodiment of the present invention, the preset loss function includes a comparison loss function and a classification loss function;
wherein the contrast loss function is represented by formula (9):
（9），
wherein the classification loss function is represented by formula (10):
（10）,
wherein,indicating temperature super parameter, ">A positive pair of samples is represented and,/>representing a sample image currently being trained, +.>Representing the number of training sample images, +.>And->Different hyper-parameters representing the classification loss function, < +.>Truth label representing training sample image +.>Representing the predicted value of the real-time fall detection model.
According to a third aspect of the present invention, there is provided an electronic device comprising:
one or more processors;
storage means for storing one or more programs,
wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform a method of real-time fall detection based on millimeter wave radar and a training method of a real-time fall detection model.
According to a fourth aspect of the invention, there is provided a computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform a method of real-time fall detection based on millimeter wave radar and a training method of a real-time fall detection model.
According to the real-time falling detection method based on the millimeter wave radar, according to the preset signal processing method, the millimeter wave radar is used for processing the transmitting signals and the receiving signals according to the time sequence so as to filter the influence of environmental noise, and the intermediate frequency signal with the environmental noise filtered is processed by the real-time falling detection model obtained through comparison and learning training, so that the falling detection of the target to be detected in real time is realized. The real-time falling detection method provided by the invention is not influenced by the environment of the target to be detected, and expands the application scene of the real-time falling detection method on the premise of improving the accuracy of the real-time falling detection.
Drawings
Fig. 1 is a flowchart of a method of real-time fall detection based on millimeter wave radar according to an embodiment of the invention;
fig. 2 is a flow chart of a training method of a real-time fall detection model according to an embodiment of the invention;
FIG. 3 is a schematic diagram of a data acquisition environment according to an embodiment of the invention;
fig. 4 is a schematic diagram of a sequence of distance and angle diagrams when an object to be detected is in a falling state according to an embodiment of the present invention;
fig. 5 is a schematic diagram of a neural network architecture of a real-time fall detection model according to an embodiment of the invention;
FIG. 6 is a functional schematic of a false positive suppression module according to an embodiment of the present invention;
fig. 7 is a block diagram of an electronic device adapted to implement a real-time fall detection method based on millimeter wave radar and a training method of a real-time fall detection model, according to an embodiment of the invention.
Detailed Description
The present invention will be further described in detail below with reference to specific embodiments and with reference to the accompanying drawings, in order to make the objects, technical solutions and advantages of the present invention more apparent.
In order to solve the technical problems of the prior art, the invention provides a real-time falling detection method based on a millimeter wave radar, which is characterized in that an intermediate frequency signal is obtained by mixing a millimeter wave radar transmitting signal and a millimeter wave radar receiving signal, an image superposition result required by a real-time falling detection model is obtained based on the intermediate frequency signal, and the accuracy of falling detection for the purpose of detection can be improved and the application scene is expanded by combining with the real-time falling detection model which is trained.
It should be specifically noted that, in the technical scheme disclosed by the invention, the acquisition of the related object character data to be detected obtains the authorization of the related party, and the data is processed, applied and stored under the permission of the related party, so that the related process accords with the rules of laws and regulations, necessary and reliable confidentiality measures are adopted, and the requirements of popular regulations are met.
Fig. 1 is a flowchart of a real-time fall detection method based on millimeter wave radar according to an embodiment of the invention.
As shown in fig. 1, the method for detecting a real-time fall based on millimeter wave radar includes operations S110 to S140.
In operation S110, a mixing process is performed on a transmission signal and a reception signal of the millimeter wave radar, to obtain an intermediate frequency signal, where the reception signal is a radar echo signal reflected by a target to be detected.
Optionally, the millimeter wave radar is prevented from being 2-3 meters above the to-be-detected area where the to-be-detected target is located, and the to-be-detected target is not more than 7 meters below the radar at the most.
In operation S120, the intermediate frequency signal is respectively subjected to fast fourier transform along a distance dimension, a speed dimension, and an angle dimension to obtain distance information, speed information, and angle information, and a distance-doppler-angle three-dimensional matrix is obtained based on the distance information, the speed information, and the angle information.
It will be clear to those skilled in the art that the doppler frequency and velocity information are essentially the same.
In operation S130, based on a preset weighted summation formula, the distance-doppler-angle three-dimensional matrix is weighted and summed in the velocity dimension to obtain a weighted distance-angle heat map, and the weighted distance-angle heat maps of consecutive multiframes are subjected to image superposition in the channel dimension to obtain an image superposition result.
According to an embodiment of the present invention, the above distance information is represented by formula (1):
（1），
wherein the speed information is represented by formula (2):
（2），
wherein the angle information is represented by formula (3):
（3），
wherein the preset weighted sum formula is represented by formula (4):
（4），
wherein,characterizing the frequency of the intermediate frequency signal, ">Characterizing the slope of the intermediate frequency signal->Characterizing the speed of light->Characterizing the wavelength of the intermediate frequency signal, ">Characterizing the phase difference between two adjacent intermediate frequency signals, a ∈>Characterizing the time difference between two adjacent intermediate frequency signals, a ∈>Characterizing the distance between two adjacent receive antennas of a millimeter wave radar,/->Representing the intermediate frequency after fast fourier transformPower of signal>And->Respectively representing different weight super parameters related to the speed information.
Wherein,，/>the determination can be made by specific experiments.
In operation S140, the image superposition result is segmented according to the preset segmentation rule, and the image segmentation result is processed by using the trained real-time fall detection model according to the false alarm suppression mechanism, so as to obtain the fall detection result of the target to be detected.
Optionally, the preset segmentation rule segments the image superposition result by using a sliding window with a duration of 2 seconds once every 0.4 seconds.
According to the embodiment of the invention, the false alarm suppression mechanism indicates that the object to be detected falls when a plurality of falls are detected in a plurality of continuous time windows, and the object to be detected falls when a plurality of falls are not detected in a plurality of continuous time windows, so that the object to be detected falls into false alarm;
wherein the probability of false alarm is represented by formula (5):
（5），
wherein,classification result output by the real-time fall detection model representing training completion,/->Indicating the time-of-day position of the time window in which the object to be detected falls.
Optionally, the false positive suppression mechanism is that more than two falls are detected in four consecutive windows, otherwise the falls are considered false positive.
According to the real-time falling detection method based on the millimeter wave radar, according to the preset signal processing method, the millimeter wave radar is used for processing the transmitting signals and the receiving signals according to the time sequence so as to filter the influence of environmental noise, and the intermediate frequency signals with the environmental noise filtered are processed by training the real-time falling detection model through the comparison learning method, so that the falling detection of the target to be detected is realized in real time. The real-time falling detection method provided by the invention is not influenced by the environment of the target to be detected, and expands the application scene of the real-time falling detection method on the premise of improving the accuracy of the real-time falling detection.
Fig. 2 is a flowchart of a training method of a real-time fall detection model according to an embodiment of the invention.
As shown in fig. 2, the training method of the real-time fall detection model is applied to a real-time fall detection method based on millimeter wave radar, and includes operations S210 to S240.
In operation S210, a training transmitting signal and a training receiving signal of the millimeter wave radar are preprocessed to obtain a training sample image, and the training sample image is subjected to data enhancement to obtain an enhanced training sample image, wherein the training receiving signal is a radar echo signal reflected by a target to be detected.
According to an embodiment of the present invention, preprocessing the training transmission signal and the training reception signal of the millimeter wave radar to obtain a training sample image includes: mixing a training transmitting signal and a training receiving signal of the millimeter wave radar to obtain a training intermediate frequency signal, and respectively performing fast Fourier transformation on the training intermediate frequency signal along a speed dimension, a distance dimension and an angle dimension to obtain speed information, distance information and angle information; and obtaining a distance-Doppler-angle three-dimensional matrix based on the speed information, the distance information and the angle information, carrying out weighted summation on the distance-Doppler-angle three-dimensional matrix in the speed dimension, and carrying out image superposition on the obtained weighted distance-angle heat map of the continuous multiframes in the channel dimension to obtain a training sample image.
In operation S220, the enhanced training sample image is processed by a signal processing method to construct a positive sample pair with a truth value tag, and the positive sample pair is processed by a real-time fall detection model to obtain a detection result, wherein the real-time fall detection model is constructed based on a contrast learning neural network and a classification neural network.
According to an embodiment of the present invention, the data enhancing the training sample image to obtain an enhanced training sample image includes: carrying out data enhancement on the training sample image through image inversion operation, image translation operation and image frame extraction operation to obtain a data enhanced training sample image; and carrying out data enhancement on the angle dimension on the training sample image with the data enhancement through a re-weighting operation, so as to obtain an enhanced training sample image.
According to the embodiment of the invention, the data enhancement is performed on the data-enhanced training sample image in the angle dimension through the re-weighting operation, so that the enhanced training sample image comprises angle resolution data enhancement, signal strength data enhancement, data enhancement of vertical tangential angle offset and horizontal tangential angle offset;
wherein the angular resolution data enhancement is represented by formula (6):
（6），
wherein the signal strength data enhancement is represented by formula (7):
（7），
wherein the data enhancement of the vertical tangential angle offset and the horizontal tangential angle offset is represented by formula (8):
（8），
wherein,characterizing the wavelength of the intermediate frequency signal, ">Representing the distance between adjacent antennas of a millimeter wave radar, +.>Representing the included angle between the object to be detected and the right direction of the millimeter wave radar, and the angle is +>Representing the signal power of the millimeter wave radar transmit antenna,represents the gain of the millimeter wave radar transmitting antenna, +.>Represents the gain of the millimeter wave radar receiving antenna, +.>Representing the wavelength of the millimeter wave radar transmission signal, +.>Representing the number of antennas of a millimeter wave radar, +.>And->Representing coordinates of the millimeter wave radar in a three-dimensional space; />Representing the scattering cross section of a millimeter wave radar by adjusting +.>Can achieve data enhancement in the angular dimension, +.>Expressed in (/ -)>) Center of rotation pair->Angle of rotation, (-)>) Representing the original spatial coordinates of the object to be detected, +.>Representing the spatial coordinates of the object to be detected after rotation, < > and>representing the distance between the object to be detected and the millimeter wave radar.
In operation S230, the detection result and the truth label of the positive sample pair are processed by using a preset loss function to obtain a loss value, and the real-time fall detection model is optimized and updated according to the loss value.
According to an embodiment of the present invention, the preset loss function includes a comparison loss function and a classification loss function;
wherein the contrast loss function is represented by formula (9):
（9），
wherein the classification loss function is represented by formula (10):
（10），
wherein,indicating temperature super parameter, ">Representing positive sample pairs, ++>Representing a sample image currently being trained, +.>Representing the number of training sample images, +.>And->Different hyper-parameters representing the classification loss function, < +.>Truth label representing training sample image +.>Representing the predicted value of the real-time fall detection model.
Alternatively, the process may be carried out in a single-stage,，/>。
in operation S240, the operations of data preprocessing, data enhancement, positive sample pair construction, model processing, loss calculation, and parameter optimization and updating are iterated until the preset training condition is satisfied, thereby obtaining a real-time fall detection model after training is completed.
The method for detecting the real-time falling based on the millimeter wave radar and the training method for the real-time falling detection model provided by the invention are further described in detail through the specific embodiments and the accompanying drawings.
FIG. 3 is a schematic diagram of a data acquisition environment according to an embodiment of the invention.
Fig. 4 is a schematic diagram of a sequence of distance and angle diagrams when an object to be detected is in a falling state according to an embodiment of the present invention.
Fig. 5 is a schematic diagram of a neural network architecture of a real-time fall detection model according to an embodiment of the invention.
Fig. 6 is a functional schematic diagram of a false positive suppression module according to an embodiment of the present invention.
In the data acquisition and processing module, an environment in which relevant radar data of an object to be detected is acquired is shown in fig. 3. Firstly, a millimeter wave radar is arranged at a position 2-3 meters above a region to be detected, the detection passing area is about 4 square meters to 6 square meters, and the target to be detected is not more than 7 meters below the radar at the most, and a plurality of linear frequency modulation signals which are arranged in time sequence and received corresponding signals are sent by the millimeter wave radar for processing; secondly, mixing a transmitting signal and a receiving signal to obtain an intermediate frequency signal, performing Fast Fourier Transform (FFT) on the intermediate frequency signal along a distance dimension, a speed dimension and an angle dimension to obtain corresponding distance, speed and angle information, and obtaining a distance-Doppler-angle three-dimensional matrix; finally, the three-dimensional matrix is weighted and summed in the speed dimension to obtain a weighted distance angle heat map, and the distance angle maps of forty frames of 2 seconds are overlapped to be used as the input of a neural network, as shown in fig. 4, when the object to be detected is in a falling state, the weighted distance angle time sequence heat map is shown in fig. 4, the horizontal axis of each map is an angle, the vertical axis is a distance, and the value of each point is the weighted sum of the signal intensity and the speed of the point.
The above distance information is represented by formula (1):
（1），
wherein the speed information is represented by formula (2):
（2），
wherein the angle information is represented by formula (3):
（3），
wherein the preset weighted sum formula is represented by formula (4):
（4），
wherein,characterizing the frequency of the intermediate frequency signal, ">Characterizing the slope of the intermediate frequency signal->Characterizing the speed of light->Characterizing the wavelength of the intermediate frequency signal, ">Characterizing the phase difference between two adjacent intermediate frequency signals, a ∈>Characterizing the time difference between two adjacent intermediate frequency signals, a ∈>Characterizing the distance between two adjacent receive antennas of a millimeter wave radar,/->Representing the power of the intermediate frequency signal after fast fourier transformation,/and the like>And->Respectively representing different weight super parameters related to the speed information.
In the contrast learning module, data samples for training the real-time fall detection model are input into the real-time fall detection model to obtain the output classification. The neural network architecture of the real-time fall detection model is shown in fig. 5, and the backbone network of the neural network architecture comprises at least three 3D convolution layers and at least two full connection layers, and the fall detection model is trained by using a contrast learning method. Positive sample pairs are constructed by signal processing methods provided at the data acquisition and processing module. And generating a positive sample pair of the samples through data enhancement, encoding through a backbone network to obtain characteristics, and calculating the contrast loss between the two characteristics. And cross entropy classification loss is calculated between the true value and the output. The two losses are co-optimized in the network, and the loss functions for training the real-time fall detection model include a contrast loss function and a classification loss function.
Wherein the contrast loss function is represented by formula (9):
（9），
wherein the classification loss function is represented by formula (10):
（10），
wherein,indicating temperature super parameter, ">Representing positive sample pairs, ++>Representing a sample image currently being trained, +.>Representing the number of training sample images, +.>And->Different hyper-parameters representing the classification loss function, < +.>Truth label representing training sample image +.>Representing the predicted value of the real-time fall detection model.
The input training sample image is subjected to operations such as inversion, translation, frame extraction and the like to carry out data enhancement, and meanwhile, the angle can be re-weighted to carry out data enhancement while the data processing is carried out.
Wherein the angular resolution data enhancement is represented by formula (6):
（6），
wherein the signal strength data enhancement is represented by formula (7):
（7），
wherein the data enhancement of the vertical tangential angle offset and the horizontal tangential angle offset is represented by formula (8):
（8），
wherein,characterizing the wavelength of the intermediate frequency signal, ">Representing the distance between adjacent antennas of a millimeter wave radar, +.>Representing the included angle between the object to be detected and the right direction of the millimeter wave radar, and the angle is +>Representing the signal power of the millimeter wave radar transmit antenna,represents the gain of the millimeter wave radar transmitting antenna, +.>Represents the gain of the millimeter wave radar receiving antenna, +.>Representing the wavelength of the millimeter wave radar transmission signal, +.>Representing the number of antennas of a millimeter wave radar, +.>And->Representing coordinates of the millimeter wave radar in a three-dimensional space; />Representing the scattering cross section of a millimeter wave radar by adjusting +.>Can achieve data enhancement in the angular dimension, +.>Expressed in (/ -)>) Center of rotation pair->Angle of rotation, (-)>) Representing the original spatial coordinates of the object to be detected, +.>Representing the spatial coordinates of the object to be detected after rotation, < > and>representing the distance between the object to be detected and the millimeter wave radar.
Angle weighted data enhancement: adjusting in summationIs the angle cosine value.
As shown in fig. 6, the present invention dynamically balances false positives, i.e. if more than one fall is detected for four consecutive time windows, a fall is acknowledged, otherwise it will be ignored as a false positive. In the real-time detection module, a sliding window with the time length of 2s is used for dividing once every 0.4s, the divided data are input into a trained neural network for classification, and whether the user falls down or not is output through a false alarm suppression mechanism, so that real-time detection can be realized; the false positive suppression is specifically expressed as detecting more than two falls within four consecutive windows, otherwise the falls are considered as false positive.
Wherein the probability of false alarm is represented by formula (5):
（5），
wherein,classification result output by the real-time fall detection model representing training completion,/->Indicating the time-of-day position of the time window in which the object to be detected falls.
For independent falling samples, the probability of false alarm can be effectively reduced on the basis of not reducing the recall rate through detection and suppression of a plurality of windows.
According to the specific embodiment of the invention, through the processing of millimeter wave radar signals, the performance of the system in different environments is effectively improved through weighting and data enhancement; the training precision is improved by a contrast learning method, and the practicability in daily life is improved by a method for detecting false positive inhibition in real time.
The invention uses the linear frequency modulation wave radar, has no great limitation on the placement mode and the position of the radar, and outputs real-time falling detection results through the signal processing module, the contrast learning module and the false alarm suppression module. Each module in the invention can run on the CPU of the computer in real time, and can be deployed on a smaller platform. And the lightweight real-time falling detection is achieved.
Fig. 7 schematically shows a block diagram of an electronic device adapted to implement a real-time fall detection method based on millimeter wave radar and a training method of a real-time fall detection model, according to an embodiment of the invention.
As shown in fig. 7, an electronic device 700 according to an embodiment of the present invention includes a processor 701 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 702 or a program loaded from a storage section 708 into a Random Access Memory (RAM) 703. The processor 701 may include, for example, a general purpose microprocessor (e.g., a CPU), an instruction set processor and/or an associated chipset and/or a special purpose microprocessor (e.g., an Application Specific Integrated Circuit (ASIC)), or the like. The processor 701 may also include on-board memory for caching purposes. The processor 701 may comprise a single processing unit or a plurality of processing units for performing different actions of the method flow according to an embodiment of the invention.
In the RAM 703, various programs and data necessary for the operation of the electronic apparatus 700 are stored. The processor 701, the ROM 702, and the RAM 703 are connected to each other through a bus 704. The processor 701 performs various operations of the method flow according to an embodiment of the present invention by executing programs in the ROM 702 and/or the RAM 703. Note that the program may be stored in one or more memories other than the ROM 702 and the RAM 703. The processor 701 may also perform various operations of the method flow according to embodiments of the present invention by executing programs stored in one or more memories.
According to an embodiment of the invention, the electronic device 700 may further comprise an input/output (I/O) interface 705, the input/output (I/O) interface 705 also being connected to the bus 704. The electronic device 700 may also include one or more of the following components connected to the I/O interface 705: an input section 706 including a keyboard, a mouse, and the like; an output portion 707 including a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, a speaker, and the like; a storage section 708 including a hard disk or the like; and a communication section 709 including a network interface card such as a LAN card, a modem, or the like. The communication section 709 performs communication processing via a network such as the internet. The drive 710 is also connected to the I/O interface 705 as needed. A removable medium 711 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 710 as necessary, so that a computer program read therefrom is mounted into the storage section 708 as necessary.
The present invention also provides a computer-readable storage medium that may be embodied in the apparatus/device/system described in the above embodiments; or may exist alone without being assembled into the apparatus/device/system. The computer-readable storage medium carries one or more programs which, when executed, implement methods in accordance with embodiments of the present invention.
According to embodiments of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example, but is not limited to: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. For example, according to an embodiment of the invention, the computer-readable storage medium may include ROM 702 and/or RAM 703 and/or one or more memories other than ROM 702 and RAM 703 described above.
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The foregoing embodiments have been provided for the purpose of illustrating the general principles of the present invention, and are not meant to limit the scope of the invention, but to limit the invention thereto.
Claims (7)
1. The real-time fall detection method based on the millimeter wave radar is characterized by comprising the following steps of:
mixing the transmitting signal and the receiving signal of the millimeter wave radar to obtain an intermediate frequency signal, wherein the receiving signal is a radar echo signal reflected by a target to be detected;
performing fast Fourier transform on the intermediate frequency signals along a distance dimension, a speed dimension and an angle dimension respectively to obtain distance information, speed information and angle information respectively, and obtaining a distance-Doppler-angle three-dimensional matrix based on the distance information, the speed information and the angle information;
based on a preset weighted summation formula, carrying out weighted summation on the distance-Doppler-angle three-dimensional matrix in the speed dimension to obtain a weighted distance angle heat map, and carrying out image superposition on the weighted distance angle heat map of continuous multiframes in the channel dimension to obtain an image superposition result;
dividing the image superposition result according to a preset dividing rule, and processing the image division result by using a real-time fall detection model after training according to a false alarm suppression mechanism to obtain a fall detection result of the target to be detected;
the real-time fall detection model after training is obtained through training by the following contrast learning method:
preprocessing a training transmitting signal and a training receiving signal of the millimeter wave radar to obtain a training sample image, and carrying out data enhancement on the training sample image to obtain an enhanced training sample image, wherein the training receiving signal is a radar echo signal reflected by a target to be detected;
processing the enhanced training sample image through a signal processing method to construct a positive sample pair with a truth value label, and processing the positive sample pair by utilizing the real-time falling detection model to obtain a detection result, wherein the real-time falling detection model is constructed based on a comparison learning neural network and a classification neural network;
processing the detection result and the truth value label of the positive sample pair by using a preset loss function to obtain a loss value, and carrying out parameter optimization and updating on the real-time falling detection model according to the loss value;
iterative data preprocessing, data enhancement, positive sample pair construction, model processing, loss calculation and parameter optimization and updating operations are carried out until preset training conditions are met, so that a real-time fall detection model with completed training is obtained;
the step of data enhancement of the training sample image, the step of obtaining the enhanced training sample image comprises the following steps:
carrying out data enhancement on the training sample image through image inversion operation, image translation operation and image frame extraction operation to obtain a data enhanced training sample image;
carrying out data enhancement on the data enhanced training sample image in the angle dimension through a re-weighting operation to obtain the enhanced training sample image;
carrying out data enhancement on the angle dimension on the data enhanced training sample image through a re-weighting operation, and obtaining the enhanced training sample image which comprises angle resolution data enhancement, signal strength data enhancement, vertical tangential angle offset and horizontal tangential angle offset data enhancement;
wherein the angular resolution data enhancement is represented by equation (6):
（6），
wherein the signal strength data enhancement is represented by formula (7):
（7），
wherein the data enhancement of the vertical tangential angle offset and the horizontal tangential angle offset is represented by formula (8):
（8），
wherein,characterizing the wavelength of said intermediate frequency signal, +.>Represents the distance between adjacent antennas of the millimeter wave radar, < >>Representing the included angle between the object to be detected and the opposite direction of the millimeter wave radar, and allowing the object to be detected to be +.>Representing the signal power of the millimeter wave radar transmitting antenna, < >>Represents the gain of the millimeter wave radar transmitting antenna, < >>Representing the gain of the millimeter wave radar receiving antenna, < >>Representing the wavelength of said millimeter wave radar transmission signal, < >>Representing the number of antennas of the millimeter wave radar, +.>And->Representing coordinates of the millimeter wave radar in a three-dimensional space; />Representing the scattering cross section of the millimeter wave radar by adjusting +.>Can achieve data enhancement in the angular dimension, +.>Expressed in (/ -)>) The angle at which the centre of rotation is rotated, (-)>) Representing the original spatial coordinates of said object to be detected, < >>Representing the spatial coordinates of the object to be detected after rotation, < >>Representing the distance between the object to be detected and the millimeter wave radar.
2. The method of claim 1, wherein the distance information is represented by formula (1):
（1），
wherein the speed information is represented by formula (2):
（2），
wherein the angle information is represented by formula (3):
（3），
wherein the preset weighted sum formula is represented by formula (4):
（4），
wherein,characterizing the frequency of said intermediate frequency signal, +.>Characterizing the slope of said intermediate frequency signal, +.>Characterizing the speed of light->Characterizing the wavelength of said intermediate frequency signal, +.>Characterizing a phase difference between two adjacent intermediate frequency signals of the millimeter wave radar, < >>Characterizing the time difference between two adjacent said intermediate frequency signals,/and>characterizing the distance between two adjacent receiving antennas, < >>Representing the power of the intermediate frequency signal after fast fourier transformation,/and the like>And->Respectively representing different weight super parameters related to the speed information.
3. The method according to claim 1, wherein the false positive suppression mechanism indicates that a plurality of falls are detected within a plurality of consecutive time windows, the target to be detected is determined to be falling, and a plurality of falls are not detected within a plurality of consecutive time windows, the target to be detected is false positive;
wherein the probability of false alarm is represented by formula (5):
（5），
wherein,representing the classification result output by the trained real-time fall detection model, < >>And the time position of the time window for the falling of the target to be detected is represented.
4. The method of claim 1, wherein preprocessing the training transmit signal and the training receive signal of the millimeter wave radar to obtain a training sample image comprises:
mixing a training transmitting signal and a training receiving signal of the millimeter wave radar to obtain a training intermediate frequency signal, and respectively performing fast Fourier transformation on the training intermediate frequency signal along a speed dimension, a distance dimension and an angle dimension to obtain speed information, distance information and angle information;
and obtaining a distance-Doppler-angle three-dimensional matrix based on the speed information, the distance information and the angle information, carrying out weighted summation on the distance-Doppler-angle three-dimensional matrix in a speed dimension, and carrying out image superposition on the obtained weighted distance-angle heat map of continuous multiframes in a channel dimension to obtain a training sample image.
5. The method of claim 1, wherein the predetermined loss function comprises a contrast loss function and a classification loss function;
wherein the contrast loss function is represented by formula (9):
（9），
wherein the classification loss function is represented by formula (10):
（10），
wherein,indicating temperature super parameter, ">Representing the positive sample pair, ++>Representing a sample image currently being trained, +.>Representing the number of training sample images, +.>And->Different hyper-parameters representing said class loss function,/->Truth labels representing the training sample images, < ->Representing predicted values of the real-time fall detection model.
6. An electronic device, comprising:
one or more processors;
storage means for storing one or more programs,
wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method of any of claims 1-5.
7. A computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to perform the method according to any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311291588.6A CN117031434B (en) | 2023-10-08 | 2023-10-08 | Real-time falling detection method based on millimeter wave radar |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311291588.6A CN117031434B (en) | 2023-10-08 | 2023-10-08 | Real-time falling detection method based on millimeter wave radar |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117031434A CN117031434A (en) | 2023-11-10 |
| CN117031434B true CN117031434B (en) | 2024-02-20 |
Family
ID=88630400
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311291588.6A Active CN117031434B (en) | 2023-10-08 | 2023-10-08 | Real-time falling detection method based on millimeter wave radar |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117031434B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114004255A (en) * | 2021-10-25 | 2022-02-01 | 中国科学技术大学 | Gesture detection method, gesture detection device, electronic device, and readable storage medium |
| CN114077892A (en) * | 2020-08-14 | 2022-02-22 | ***通信有限公司研究院 | Human body skeleton sequence extraction and training method, device and storage medium |
| CN114387672A (en) * | 2022-01-18 | 2022-04-22 | 北京理工大学 | Human body behavior classification method based on time-space-frequency three-dimensional radar point cloud |
| CN115902878A (en) * | 2022-11-21 | 2023-04-04 | 广州大学 | Millimeter wave radar human behavior recognition method |
| CN115908442A (en) * | 2023-01-06 | 2023-04-04 | 山东巍然智能科技有限公司 | Image panorama segmentation method for unmanned aerial vehicle ocean monitoring and model building method |
| CN116338684A (en) * | 2023-05-29 | 2023-06-27 | 长沙莫之比智能科技有限公司 | Human body falling detection method and system based on millimeter wave radar and deep learning |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106338732B (en) * | 2016-08-23 | 2019-02-26 | 华讯方舟科技有限公司 | A kind of millimeter wave three-dimensional holographic imaging method and system |
| CN112560972B (en) * | 2020-12-21 | 2021-10-08 | 北京航空航天大学 | Target detection method based on millimeter wave radar prior positioning and visual feature fusion |
-
2023
- 2023-10-08 CN CN202311291588.6A patent/CN117031434B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114077892A (en) * | 2020-08-14 | 2022-02-22 | ***通信有限公司研究院 | Human body skeleton sequence extraction and training method, device and storage medium |
| CN114004255A (en) * | 2021-10-25 | 2022-02-01 | 中国科学技术大学 | Gesture detection method, gesture detection device, electronic device, and readable storage medium |
| CN114387672A (en) * | 2022-01-18 | 2022-04-22 | 北京理工大学 | Human body behavior classification method based on time-space-frequency three-dimensional radar point cloud |
| CN115902878A (en) * | 2022-11-21 | 2023-04-04 | 广州大学 | Millimeter wave radar human behavior recognition method |
| CN115908442A (en) * | 2023-01-06 | 2023-04-04 | 山东巍然智能科技有限公司 | Image panorama segmentation method for unmanned aerial vehicle ocean monitoring and model building method |
| CN116338684A (en) * | 2023-05-29 | 2023-06-27 | 长沙莫之比智能科技有限公司 | Human body falling detection method and system based on millimeter wave radar and deep learning |
Non-Patent Citations (3)
| Title |
|---|
| Millimeter-Wave Radar Monitoring for Elder’s Fall Based on Multi-View Parameter Fusion Estimation and Recognition;Xiang Feng 等;《remote sensing》;第15卷(第8期);1-20 * |
| 基于多通道调频连续波毫米波雷达的微动手势识别;夏朝阳 等;电子与信息学报;第42卷(第01期);164-172 * |
| 基于阵列子通道投影图的雷达图像增强算法;戴永鹏 等;《信号处理》;第39卷(第09期);1552-1561 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117031434A (en) | 2023-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3825728A1 (en) | Method and device to improve radar data using reference data background | |
| CN101676744B (en) | Method for tracking small target with high precision under complex background and low signal-to-noise ratio | |
| CN110007366B (en) | Life searching method and system based on multi-sensor fusion | |
| Ahmadi et al. | Small dim object tracking using frequency and spatial domain information | |
| CN108010065A (en) | Low target quick determination method and device, storage medium and electric terminal | |
| CN201114551Y (en) | Multi-mode and multi-target precise tracking device | |
| CN115343704A (en) | Gesture recognition method of FMCW millimeter wave radar based on multi-task learning | |
| CN112766108A (en) | SAR image target detection method based on context information | |
| CN116129371A (en) | Traffic target detection method and device and electronic equipment | |
| CN117031434B (en) | Real-time falling detection method based on millimeter wave radar | |
| CN114004255A (en) | Gesture detection method, gesture detection device, electronic device, and readable storage medium | |
| CN110687513A (en) | Human body target detection method, device and storage medium | |
| Wang et al. | Interference mitigation for automotive fmcw radar based on contrastive learning with dilated convolution | |
| Danso et al. | An optimal defect recognition security-based terahertz low resolution image system using deep learning network | |
| CN113514796B (en) | Passive positioning method, system and medium | |
| CN114637001B (en) | Target detection method, device and equipment of radar and storage medium | |
| Lin et al. | Synthetic aperture radar image aircraft detection based on target spatial imaging characteristics | |
| CN113657293B (en) | Living body detection method, living body detection device, electronic equipment, medium and program product | |
| CN115131756A (en) | Target detection method and device | |
| CN113311405B (en) | Regional population counting method and device, computer equipment and storage medium | |
| CN116933002A (en) | Action detection method and device, terminal equipment and storage medium | |
| CN116702015B (en) | Signal processing method, device, unmanned aerial vehicle and computer readable storage medium | |
| CN113822372A (en) | Unmanned aerial vehicle detection method based on YOLOv5 neural network | |
| Chuang et al. | Applying an Adaptive Signal Identification Method to Improve Vessel Echo Detection and Tracking for SeaSonde HF Radar | |
| CN113935359A (en) | Vital sign detection method and vital sign detection equipment |
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 |