CN114595748A - Data segmentation method for fall protection system - Google Patents
Data segmentation method for fall protection system Download PDFInfo
- Publication number
- CN114595748A CN114595748A CN202210157764.6A CN202210157764A CN114595748A CN 114595748 A CN114595748 A CN 114595748A CN 202210157764 A CN202210157764 A CN 202210157764A CN 114595748 A CN114595748 A CN 114595748A
- Authority
- CN
- China
- Prior art keywords
- data
- data set
- protection system
- data segmentation
- training
- 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
- 238000000034 method Methods 0.000 title claims abstract description 51
- 230000011218 segmentation Effects 0.000 title claims abstract description 45
- 238000012549 training Methods 0.000 claims abstract description 31
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 25
- 238000013507 mapping Methods 0.000 claims abstract description 20
- 230000009471 action Effects 0.000 claims abstract description 15
- 238000001514 detection method Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 230000008569 process Effects 0.000 claims description 12
- 230000006399 behavior Effects 0.000 claims description 9
- 238000011176 pooling Methods 0.000 claims description 7
- 238000007477 logistic regression Methods 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 3
- 238000010801 machine learning Methods 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 3
- 239000013598 vector Substances 0.000 claims description 3
- 238000012800 visualization Methods 0.000 claims description 3
- 230000007547 defect Effects 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 4
- 230000001133 acceleration Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000005180 public health Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/2163—Partitioning the feature space
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
Abstract
The invention relates to a data segmentation method for a fall protection system, which comprises the following steps: acquiring a data set of falling actions and daily behavior actions of a user; preprocessing the acquired data set, intercepting data of an inertial sensor worn by the user before the user falls down and impacts the ground, and establishing a training data set; training a training data set based on a multi-channel convolutional neural network model MC-CNN to obtain feature maps and corresponding weights of all channels; and (3) combining the obtained feature map and corresponding weight with an importance mapping method to obtain the importance of each sequence position for data segmentation, thereby determining a specific area for data segmentation and further being used for fall detection of a fall protection system. The method can clearly divide the triggering condition of falling detection and the intercepting length of the window, avoids the defect that the length of the data dividing window is determined by experience in the existing dividing method, and solves the problem that a falling protection system is difficult to divide data.
Description
Technical Field
The invention relates to the technical field of sensor data segmentation, in particular to a data segmentation method for a fall protection system.
Background
Falling of elderly people is a major public health problem worldwide. Tumble is the leading cause of death, disability and loss of independence in the elderly over age 65, and is also the leading cause of hospitalization (one elderly is hospitalized by tumbling every 11 seconds). Therefore, an automatic fall protection system is very interesting.
The main purpose of a fall protection system is to accurately detect when a fall has occurred in order to activate the fall protection device before the body hits the ground in order to reduce fall injuries.
In general, fall protection systems are more suitable for wearable systems based on inertial sensors. The core of the fall protection system is a detection algorithm which is responsible for continuously classifying any movement of the human body as a fall or a daily life action. It has been widely verified that artificial intelligence based detection methods can achieve higher accuracy than threshold based methods.
Since for fall protection systems based on artificial intelligence models the classification features can only be extracted from data segments of limited duration, the detection algorithm first segments the data, i.e. divides the continuous data stream obtained from the inertial sensors into a plurality of data segments. There are two general approaches to data segmentation, the first being a fixed duration sliding window. The duration of the sliding window defines the boundary of the data segment, and features and classification are further extracted after interception. Although the method is simple, the calculation amount is large, and the method is not suitable for a fall protection system with a high real-time requirement. The second method first sets a trigger condition to detect a potential fall event in the input data stream by searching for inertial sensor values that exceed a preset threshold. When a potential fall event is detected, the data segments are intercepted through one or more windows, and classification is achieved. However, how the window duration is determined, there is no clear method at present, and most are experimental empirical values. If the duration of the window is too long, the real-time property of falling detection is influenced, and the protection device does not have enough time to open; if the window duration is too short, the accuracy of fall detection will be affected.
Disclosure of Invention
To address the above-discussed deficiencies and inadequacies of the prior art, it is an object of the present invention to provide a method of inertial sensor data segmentation. The data segmentation method is realized by a multi-channel convolutional neural network MC-CNN and an importance mapping method. The method comprises the steps of preprocessing an existing data set, intercepting data of an inertial sensor before a human body impacts the ground, constructing a training data set, training through an MC-CNN network, extracting model parameters, and determining the importance of each sequence position for classification by combining an importance mapping method, so that the clear range of data segmentation is determined.
In order to achieve the purpose, the invention adopts the following technical scheme:
a data segmentation method for a fall protection system, comprising the steps of:
s1, acquiring a data set of falling actions and daily behaviors of a user;
s2, preprocessing the acquired data set, intercepting data of an inertial sensor worn by the user before the user falls down and impacts the ground, and establishing a training data set;
s3, training a training data set based on a multi-channel convolutional neural network model MC-CNN to obtain feature maps and corresponding weights of all channels;
s4, combining the obtained feature map and corresponding weight with an importance mapping method to obtain the importance of each sequence position for data segmentation, thereby determining a specific area of the data segmentation;
and S5, setting a corresponding data segmentation algorithm in the falling protection system according to the determined data segmentation area, wherein the corresponding data segmentation algorithm is used for falling detection of the falling protection system.
In step S2, the acquired data set is preprocessed, and the preprocessing process is as follows:
s21, calculating signal amplitude vectors of three-axis inertial sensorWherein the content of the first and second substances, andrepresenting x, y and z inertial sensor values, respectively;
s22, establishing a data set: for the falling action, the SMV peak point represents the moment when the body impacts the ground, so that a signal 1s before the maximum peak point is intercepted and used as a training data set; for the daily behavior actions such as walking slowly, running, going upstairs and going downstairs, data are intercepted at the time interval of 1s and serve as a training data set.
The inertial sensor in step S2 includes a three-axis accelerometer and a gyroscope.
In step S3, the training data set is trained based on the multi-channel convolutional neural network model MC-CNN, and the training process is:
firstly, processing signals of a three-axis accelerometer and a gyroscope independently, comprehensively considering the influence of the signals of the three-axis accelerometer and the gyroscope on classification after convolution operation, merging feature maps of two parts to obtain an area which is more important for classification, and finally judging the classification quality by using a cross entropy loss function to finish the training of a data set;
in the above process, the Convolution operation is composed of four one-dimensional Convolution 1D Convolution and global average pooling layers GAP, a linear full-link layer FC, and a softmax logistic regression layer, Convolution kernels of the four one-dimensional Convolution 1D Convolution are respectively 8, 16, 32, and 64, and the global average pooling layer GAP reduces a feature mapping dimension of each channel of the fourth Convolution layer from (1 × 100) to (1 × 1).
In step S4, the importance of each sequence position for data segmentation is obtained by combining the importance mapping method, and the specific process is as follows:
after the MC-CNN model is trained, feature maps and corresponding weights of all channels can be obtained, and input inertia is subjected toSensor time series in Sk(x) Representing the output sequence on channel k, x represents the time position on the sequence,representing the weight of each channel feature k to a different class c, the input to the softmax logistic regression layer is denoted gcThen, there are:
from this, an importance mapping is built from the sequences to each class C, defined as Mc;
Mc(x) Indicating the importance of position x in the time series for classifying the series as c.
In step S4, M is implementedc(x) After the visualization, specific areas of data segmentation are determined by summarizing the rules of the importance mapping maps of the areas, and data segments corresponding to the areas are used for behavior classification based on a machine learning model so as to verify whether the data segments are falling actions.
Compared with the prior art, the invention has the beneficial effects that:
the method is realized by a multichannel convolutional neural network MC-CNN and an importance mapping method. The method comprises the steps of intercepting data of an inertial sensor before a human body impacts the ground through preprocessing of an existing data set, establishing a training data set, then training the data set through an MC-CNN network, extracting model parameters, and determining the importance of each sequence position for classification by combining an importance mapping method, so that the clear range of data segmentation is determined.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
Fig. 1 is a general flow chart of a data segmentation method for a fall protection system according to the present invention;
figure 2 is a diagram of fall signature analysis in an embodiment of the invention;
FIG. 3 is a schematic diagram of the components of a multi-channel convolutional neural network in an embodiment of the present invention;
FIG. 4 is a sample graph of an importance map in an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without any inventive step, are within the scope of the present invention.
Example (b): see fig. 1-4.
As shown in fig. 1, the data segmentation method for a fall protection system of the present invention includes the following steps:
s1, acquiring a data set of falling actions and daily behaviors of a user;
s2, preprocessing the acquired data set, intercepting data of an inertial sensor worn by the user before the user falls down and impacts the ground, and establishing a training data set;
s3, training a training data set based on a multi-channel convolutional neural network model MC-CNN to obtain feature maps and corresponding weights of all channels;
s4, combining the obtained feature map and corresponding weight with an importance mapping method to obtain the importance of each sequence position for data segmentation, thereby determining a specific area of the data segmentation;
and S5, setting a corresponding data segmentation algorithm in the falling protection system according to the determined data segmentation area, wherein the corresponding data segmentation algorithm is used for falling detection of the falling protection system.
In step S2, the acquired data set is preprocessed, and the preprocessing process is as follows:
s21, calculating signal amplitude vectors of three-axis inertial sensorWherein the content of the first and second substances, andrepresenting x, y and z inertial sensor values, respectively;
s22, establishing a data set: as shown in fig. 2, for a falling action, the SMV peak point represents the moment when the body hits the ground, so the signal 1s before the maximum peak point is intercepted as a training data set; for the daily behavior actions such as walking slowly, running, going upstairs and going downstairs, data are intercepted at the time interval of 1s and serve as a training data set.
The inertial sensor in step S2 includes a three-axis accelerometer and a gyroscope.
In step S3, the training data set is trained based on the multi-channel convolutional neural network model MC-CNN, and the training process is:
as shown in fig. 3, firstly, signals of the triaxial accelerometer and the gyroscope are processed independently, after convolution operation, the influence of the signals of the triaxial accelerometer and the gyroscope on classification is considered comprehensively, then feature maps of the two parts are merged to obtain an area which is important for classification, and finally, the classification quality is judged by using a cross entropy loss function to complete the training of a data set;
in the above process, the Convolution operation is composed of four one-dimensional Convolution 1D Convolution and global average pooling layers GAP, a linear full-link layer FC, and a softmax logistic regression layer, Convolution kernels of the four one-dimensional Convolution 1D Convolution are respectively 8, 16, 32, and 64, and the global average pooling layer GAP reduces a feature mapping dimension of each channel of the fourth Convolution layer from (1 × 100) to (1 × 1); different from the existing CNN network, the method cancels any pooling operation after the convolutional layer, thereby ensuring that the data length of the feature map is unchanged so as to be convenient for determining the subsequent segmentation area.
In step S4, the importance of each sequence position for data segmentation is obtained by combining the importance mapping method, and the specific process is as follows:
after the MC-CNN model is trained, feature maps and corresponding weights of all channels can be obtained, and S is used for an input inertial sensor time sequencek(x) Representing the output sequence on channel k, x represents the time position on the sequence,representing the weight of each channel feature k to a different class c, the input to the softmax logistic regression layer is denoted gcThen, there are:
from this, an importance mapping is built from the sequences to each class C, defined as Mc;
Mc(x) Indicating the importance of position x in the time series for classifying the series as c.
In step S4, M is implementedc(x) After the visualization, as shown in fig. 4, a specific area of data segmentation is determined by summarizing the rules of the importance mapping diagram, the darker the color bars in the diagram indicate that the corresponding position sequence is more important for classification, in this embodiment, when the acceleration count value is less than 0.93 times of the gravitational acceleration, a potential fall is determined, and from this moment, accelerometer and gyroscope data segments are intercepted according to the length of a 310ms window for behavior classification based on a machine learning model, and finally, whether the fall is detected is verified.
It should be noted that the application range of the data segmentation method is not limited to fall detection, and for other types of actions in life, rules can be trained and extracted by the method according to specific conditions, so as to obtain corresponding data segmentation positions.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the structure of the present invention in any way. Any simple modification, equivalent change and modification of the above embodiments according to the technical spirit of the present invention are within the technical scope of the present invention.
Claims (6)
1. A data segmentation method for a fall protection system, comprising the steps of:
s1, acquiring a data set of falling actions and daily behaviors of a user;
s2, preprocessing the acquired data set, intercepting data of an inertial sensor worn by the user before the user falls down and impacts the ground, and establishing a training data set;
s3, training a training data set based on a multi-channel convolutional neural network model MC-CNN to obtain feature maps and corresponding weights of all channels;
s4, obtaining the importance of each sequence position for data segmentation by combining the obtained feature map and corresponding weight with an importance mapping method, thereby determining a specific area of the data segmentation;
and S5, setting a corresponding data segmentation algorithm in the falling protection system according to the determined data segmentation area, wherein the corresponding data segmentation algorithm is used for falling detection of the falling protection system.
2. A data segmentation method for a fall protection system as claimed in claim 1, characterized by the steps of
Preprocessing the acquired data set as described in S2, where the preprocessing process is as follows:
s21, calculating signal amplitude vectors of three-axis inertial sensorWherein the content of the first and second substances, andrepresenting x, y and z inertial sensor values, respectively;
s22, establishing a data set: for the falling action, the SMV peak point represents the moment when the body impacts the ground, so that a signal 1s before the maximum peak point is intercepted and used as a training data set; for the daily behavior actions such as walking slowly, running, going upstairs and going downstairs, data are intercepted at the time interval of 1s and serve as a training data set.
3. A data segmentation method for a fall protection system as claimed in claim 1, characterized by the steps of
The inertial sensor in S2 includes a three-axis accelerometer and a gyroscope.
4. The data segmentation method for a fall protection system as claimed in claim 1, wherein the training data set is trained based on the multi-channel convolutional neural network model MC-CNN in step S3, and the training process is as follows:
firstly, signals of a three-axis accelerometer and a gyroscope are independently processed, the influence of the signals of the three-axis accelerometer and the gyroscope on classification is comprehensively considered after convolution operation, then feature maps of two parts are combined to obtain an area which is important for classification, and finally, the classification quality is judged by using a cross entropy loss function to finish the training of a data set;
in the above process, the Convolution operation is composed of four one-dimensional Convolution 1D Convolution and global average pooling layers GAP, a linear full-link layer FC, and a softmax logistic regression layer, Convolution kernels of the four one-dimensional Convolution 1D Convolution are respectively 8, 16, 32, and 64, and the global average pooling layer GAP reduces a feature mapping dimension of each channel of the fourth Convolution layer from (1 × 100) to (1 × 1).
5. A data segmentation method for a fall protection system as claimed in claim 1, wherein the importance of each sequence position for data segmentation is obtained by combining the importance mapping method in step S4, and the specific process is as follows:
after the MC-CNN model is trained, feature maps and corresponding weights of all channels can be obtained, and S is used for an input inertial sensor time sequencek(x) Representing the output sequence on channel k, x represents the time position on the sequence,representing the weight of each channel feature k to a different class c, the input to the softmax logistic regression layer is denoted gcThen, there are:
from this, an importance mapping is built from the sequences to each class C, defined as Mc;
Mc(x) Indicating the importance of position x in the time series for classifying the sequence as c.
6. A data segmentation method for a fall protection system as claimed in claim 1, wherein step S4 is implemented for Mc(x) After the visualization, specific areas of data segmentation are determined by summarizing the rules of the importance mapping maps of the areas, and data segments corresponding to the areas are used for behavior classification based on a machine learning model so as to verify whether the data segments are falling actions.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210157764.6A CN114595748B (en) | 2022-02-21 | 2022-02-21 | Data segmentation method for fall protection system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210157764.6A CN114595748B (en) | 2022-02-21 | 2022-02-21 | Data segmentation method for fall protection system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114595748A true CN114595748A (en) | 2022-06-07 |
CN114595748B CN114595748B (en) | 2024-02-13 |
Family
ID=81805527
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210157764.6A Active CN114595748B (en) | 2022-02-21 | 2022-02-21 | Data segmentation method for fall protection system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114595748B (en) |
Citations (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105310696A (en) * | 2015-11-06 | 2016-02-10 | 中国科学院计算技术研究所 | Fall detection model construction method as well as corresponding fall detection method and apparatus |
CN106846729A (en) * | 2017-01-12 | 2017-06-13 | 山东大学 | A kind of fall detection method and system based on convolutional neural networks |
WO2018048561A1 (en) * | 2016-09-09 | 2018-03-15 | Qualcomm Incorporated | Devices and methods for fall detection based on phase segmentation |
US10140544B1 (en) * | 2018-04-02 | 2018-11-27 | 12 Sigma Technologies | Enhanced convolutional neural network for image segmentation |
US20190050711A1 (en) * | 2017-08-08 | 2019-02-14 | Neusoft Corporation | Method, storage medium and electronic device for detecting vehicle crashes |
CN111274954A (en) * | 2020-01-20 | 2020-06-12 | 河北工业大学 | Embedded platform real-time falling detection method based on improved attitude estimation algorithm |
CN112016619A (en) * | 2020-08-28 | 2020-12-01 | 西安科技大学 | Fall detection method based on insoles |
CA3137030A1 (en) * | 2019-05-31 | 2020-12-03 | Maryam ZIAEEFARD | Method and processing device for training a neural network |
AU2020103901A4 (en) * | 2020-12-04 | 2021-02-11 | Chongqing Normal University | Image Semantic Segmentation Method Based on Deep Full Convolutional Network and Conditional Random Field |
CN112465905A (en) * | 2019-09-06 | 2021-03-09 | 四川大学华西医院 | Characteristic brain region positioning method of magnetic resonance imaging data based on deep learning |
WO2021057810A1 (en) * | 2019-09-29 | 2021-04-01 | 深圳数字生命研究院 | Data processing method, data training method, data identifying method and device, and storage medium |
CN112766165A (en) * | 2021-01-20 | 2021-05-07 | 燕山大学 | Falling pre-judging method based on deep neural network and panoramic segmentation |
US20210173045A1 (en) * | 2015-07-17 | 2021-06-10 | Yuqian HU | Method, apparatus, and system for fall-down detection based on a wireless signal |
US20210217178A1 (en) * | 2020-01-15 | 2021-07-15 | The Regents Of The University Of California | Systems and Methods for Trainable Deep Active Contours for Image Segmentation |
US20210248761A1 (en) * | 2020-02-10 | 2021-08-12 | Hong Kong Applied Science and Technology Research Institute Company Limited | Method for image segmentation using cnn |
CN113269786A (en) * | 2021-05-19 | 2021-08-17 | 青岛理工大学 | Assembly image segmentation method and device based on deep learning and guided filtering |
WO2021212883A1 (en) * | 2020-04-20 | 2021-10-28 | 电子科技大学 | Fall detection method based on intelligent mobile terminal |
US20220012591A1 (en) * | 2020-07-08 | 2022-01-13 | UMNAI Limited | Method for detecting and mitigating bias and weakness in artificial intelligence training data and models |
WO2022007266A1 (en) * | 2020-07-08 | 2022-01-13 | 嘉楠明芯(北京)科技有限公司 | Method and apparatus for accelerating convolutional neural network |
-
2022
- 2022-02-21 CN CN202210157764.6A patent/CN114595748B/en active Active
Patent Citations (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20210173045A1 (en) * | 2015-07-17 | 2021-06-10 | Yuqian HU | Method, apparatus, and system for fall-down detection based on a wireless signal |
CN105310696A (en) * | 2015-11-06 | 2016-02-10 | 中国科学院计算技术研究所 | Fall detection model construction method as well as corresponding fall detection method and apparatus |
WO2018048561A1 (en) * | 2016-09-09 | 2018-03-15 | Qualcomm Incorporated | Devices and methods for fall detection based on phase segmentation |
CN106846729A (en) * | 2017-01-12 | 2017-06-13 | 山东大学 | A kind of fall detection method and system based on convolutional neural networks |
US20190050711A1 (en) * | 2017-08-08 | 2019-02-14 | Neusoft Corporation | Method, storage medium and electronic device for detecting vehicle crashes |
US10140544B1 (en) * | 2018-04-02 | 2018-11-27 | 12 Sigma Technologies | Enhanced convolutional neural network for image segmentation |
CA3137030A1 (en) * | 2019-05-31 | 2020-12-03 | Maryam ZIAEEFARD | Method and processing device for training a neural network |
CN112465905A (en) * | 2019-09-06 | 2021-03-09 | 四川大学华西医院 | Characteristic brain region positioning method of magnetic resonance imaging data based on deep learning |
WO2021057810A1 (en) * | 2019-09-29 | 2021-04-01 | 深圳数字生命研究院 | Data processing method, data training method, data identifying method and device, and storage medium |
US20210217178A1 (en) * | 2020-01-15 | 2021-07-15 | The Regents Of The University Of California | Systems and Methods for Trainable Deep Active Contours for Image Segmentation |
CN111274954A (en) * | 2020-01-20 | 2020-06-12 | 河北工业大学 | Embedded platform real-time falling detection method based on improved attitude estimation algorithm |
US20210248761A1 (en) * | 2020-02-10 | 2021-08-12 | Hong Kong Applied Science and Technology Research Institute Company Limited | Method for image segmentation using cnn |
WO2021212883A1 (en) * | 2020-04-20 | 2021-10-28 | 电子科技大学 | Fall detection method based on intelligent mobile terminal |
WO2022007266A1 (en) * | 2020-07-08 | 2022-01-13 | 嘉楠明芯(北京)科技有限公司 | Method and apparatus for accelerating convolutional neural network |
US20220012591A1 (en) * | 2020-07-08 | 2022-01-13 | UMNAI Limited | Method for detecting and mitigating bias and weakness in artificial intelligence training data and models |
CN112016619A (en) * | 2020-08-28 | 2020-12-01 | 西安科技大学 | Fall detection method based on insoles |
AU2020103901A4 (en) * | 2020-12-04 | 2021-02-11 | Chongqing Normal University | Image Semantic Segmentation Method Based on Deep Full Convolutional Network and Conditional Random Field |
CN112766165A (en) * | 2021-01-20 | 2021-05-07 | 燕山大学 | Falling pre-judging method based on deep neural network and panoramic segmentation |
CN113269786A (en) * | 2021-05-19 | 2021-08-17 | 青岛理工大学 | Assembly image segmentation method and device based on deep learning and guided filtering |
Non-Patent Citations (3)
Title |
---|
AKIRA JINGUJI ET AL: "Weight Sparseness for a Feature-Map-Split-CNN Toward Low-Cost Embedded FPGAs", 《IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS》 * |
刘继忠等: "基于云通讯的监护系统多生理数据传输策略", 《南昌大学学报(工科版)》, vol. 42, no. 04 * |
杨辉等: "基于弱监督学习的目标检测研究进展", 《计算机工程与应用》, vol. 57, no. 16 * |
Also Published As
Publication number | Publication date |
---|---|
CN114595748B (en) | 2024-02-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106846729B (en) | Tumble detection method and system based on convolutional neural network | |
CN105310696B (en) | A kind of fall detection model building method and corresponding fall detection method and device | |
CN107909027B (en) | Rapid human body target detection method with shielding treatment | |
CN108509897A (en) | A kind of human posture recognition method and system | |
CN110781723B (en) | Group abnormal behavior identification method | |
CN106875630B (en) | A kind of wearable fall detection method and system based on hierarchical classification | |
WO2021115066A1 (en) | Inertial sensor-based gait feature extraction and gait recognition method | |
CN108985204A (en) | Pedestrian detection tracking and device | |
Gjoreski et al. | Accelerometer data preparation for activity recognition | |
CN104586402B (en) | A kind of feature extracting method of physical activity | |
CN112464738A (en) | Improved naive Bayes algorithm user behavior identification method based on mobile phone sensor | |
CN109271889B (en) | Action recognition method based on double-layer LSTM neural network | |
CN114881079A (en) | Human body movement intention abnormity detection method and system for wearable sensor | |
CN111603169A (en) | Pedestrian gait recognition method based on MEMS inertial sensor | |
CN114595748A (en) | Data segmentation method for fall protection system | |
CN112801000A (en) | Household old man falling detection method and system based on multi-feature fusion | |
CN109685111A (en) | Action identification method, computing system, smart machine and storage medium | |
CN114327045A (en) | Fall detection method and system based on category unbalanced signals | |
KR20230139666A (en) | Devices and Methods for Recognizing Drowsy Driving based on Classification of Drowsy Driving Types and Ensemble Deep Learning | |
CN115886794A (en) | Multi-classifier selection method, frozen gait detection system, device and storage medium | |
CN114120370A (en) | CNN-LSTM-based human body falling detection implementation method and system | |
Silva et al. | Human activity classification with inertial sensors | |
CN114916928B (en) | Human body posture multichannel convolutional neural network detection method | |
Anakala et al. | Fall Detection and Elderly Monitoring System Using the CNN | |
CN111710129B (en) | Real-time pre-collision falling detection method for old people |
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 |