CN116740813A - Analysis system and method based on AI image recognition behavior monitoring - Google Patents

Analysis system and method based on AI image recognition behavior monitoring Download PDF

Info

Publication number
CN116740813A
CN116740813A CN202310739120.2A CN202310739120A CN116740813A CN 116740813 A CN116740813 A CN 116740813A CN 202310739120 A CN202310739120 A CN 202310739120A CN 116740813 A CN116740813 A CN 116740813A
Authority
CN
China
Prior art keywords
behavior
recognition
human body
data
behaviors
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
Application number
CN202310739120.2A
Other languages
Chinese (zh)
Other versions
CN116740813B (en
Inventor
彭孔涛
李文华
曹忠文
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Videostrong Technology Co ltd
Original Assignee
Shenzhen Videostrong Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shenzhen Videostrong Technology Co ltd filed Critical Shenzhen Videostrong Technology Co ltd
Priority to CN202310739120.2A priority Critical patent/CN116740813B/en
Publication of CN116740813A publication Critical patent/CN116740813A/en
Application granted granted Critical
Publication of CN116740813B publication Critical patent/CN116740813B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/20—Movements or behaviour, e.g. gesture recognition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • 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
    • G06V10/77—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
    • G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
    • G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Human Computer Interaction (AREA)
  • Social Psychology (AREA)
  • Psychiatry (AREA)
  • Image Analysis (AREA)

Abstract

The invention relates to the technical field of AI, and provides an analysis system and a method based on AI image recognition behavior monitoring, wherein the system comprises an AI behavior recognition device, a human body behavior recognition device and a log behavior recognition device, and the AI behavior recognition device and the log behavior recognition are respectively connected with the human body behavior recognition device; the AI behavior recognition device comprises an AI image recognition device and an AI behavior recognition model; the human body behavior recognition device comprises a human body data recognition device and a first behavior recognition model; the log behavior recognition device comprises a log data acquisition device and a second behavior recognition model; acquiring AI human behaviors based on the AI behavior recognition device; obtaining model human body behaviors based on the human body behavior recognition device; obtaining log human body behaviors based on the log behavior recognition device; and monitoring and analyzing the user behavior of the target user. The analysis system based on AI image recognition behavior monitoring provided by the invention realizes accurate behavior monitoring analysis on user behaviors.

Description

Analysis system and method based on AI image recognition behavior monitoring
Technical Field
The invention relates to the technical field of AI (advanced technology attachment), in particular to an analysis system and a method based on AI image recognition behavior monitoring.
Background
The existing user behavior monitoring and analyzing method is to input the collected user images into a single behavior recognition model, recognize the behaviors of the user through the single behavior recognition model, and detect and analyze the behaviors of the user. However, in the course of behavior recognition analysis, the recognition methods and recognition accuracy of the individual recognition models are different, so that deviation occurs in behavior monitoring analysis, and thus accurate behavior monitoring analysis cannot be performed on the behavior of the user.
Disclosure of Invention
The invention provides an analysis system and a method based on AI image recognition behavior monitoring, aiming at realizing accurate behavior monitoring analysis on user behaviors.
In a first aspect, the invention provides an analysis system based on AI image recognition behavior monitoring, the system comprises an artificial intelligent AI behavior recognition device, a human body behavior recognition device and a log behavior recognition device, wherein the AI behavior recognition device is connected with the human body behavior recognition device, and the human body behavior recognition device is connected with the log behavior recognition device;
the AI behavior recognition device comprises an AI image recognition device and an AI behavior recognition model; the human body behavior recognition device comprises a human body data recognition device and a first behavior recognition model; the log behavior recognition device comprises a log data acquisition device and a second behavior recognition model;
The AI behavior recognition device is used for inputting an AI human body image obtained by recognizing the user behavior of the target user by the AI image recognition device into the AI behavior recognition model to obtain AI human body behaviors;
the human body behavior recognition device is used for inputting human body data obtained by recognizing the user behavior by the human body data recognition device into the first behavior recognition model to obtain model human body behavior;
the log behavior recognition device is used for inputting the log data obtained by the log data obtaining device according to the user behavior into the second behavior recognition model to obtain log human behaviors;
the analysis system based on AI image recognition behavior monitoring is used for monitoring and analyzing the user behavior of the target user according to the AI human behavior, the model human behavior and the log human behavior.
In a second aspect, the present invention provides an analysis method based on AI image recognition behavior monitoring, applied to the system according to the first aspect, including:
acquiring user behaviors of a target user;
performing AI identification on the user behaviors to obtain AI human body images, and inputting the AI human body images into an AI behavior identification model to obtain AI human body behaviors;
Human body recognition is carried out on the user behaviors to obtain human body data, and the human body data are input into a first behavior recognition model to obtain model human body behaviors;
acquiring log data of the user behaviors, and inputting the log data into a second behavior recognition model to obtain log human behaviors;
monitoring and analyzing the user behaviors of the target user according to the AI human behaviors, the model human behaviors and the log human behaviors;
the AI behavior recognition model is obtained by training an AI image sample and a corresponding behavior label thereof; the first behavior recognition model is obtained by training a human body data sample and a corresponding behavior label thereof; the second behavior recognition model is obtained by training a log data sample and corresponding user behavior attribute sample data for a tag label.
In one embodiment, the AI behavior recognition model includes a head recognition network, a body part recognition network, and a leg recognition network;
the step of carrying out AI identification on the user behavior to obtain an AI human body image, and inputting the AI human body image into an AI behavior identification model to obtain an AI human body behavior, comprising the following steps:
Performing AI identification on the user behaviors to obtain the behavior actions of the target user and AI human body images of human body areas;
the shoulder is used as a first segmentation characteristic line and the waist is used as a second segmentation characteristic line, the AI human body image is subjected to characteristic segmentation, and a head characteristic image, a body part characteristic image and a leg characteristic image of the target user are obtained;
inputting the head characteristic image into the head recognition network to obtain head characteristic data of the target user output by the head recognition network;
inputting the body part characteristic image into the body part recognition network to obtain body part characteristic data of the target user output by the body part recognition network;
inputting the leg characteristic image into the leg identification network to obtain leg characteristic data of the target user output by the leg identification network;
and obtaining the AI human behavior according to the head characteristic data, the body part characteristic data and the leg characteristic data of the target user.
The AI behavior recognition model further comprises a behavior recognition network and a behavior fusion network;
the obtaining the AI human behavior according to the head feature data, the body part feature data and the leg feature data of the target user includes:
Inputting the head characteristic data, the body part characteristic data and the leg characteristic data of the target user into the behavior recognition network to perform behavior recognition, so as to respectively obtain the head behavior, the body part behavior and the leg behavior of the target user;
and inputting the head behaviors, the body part behaviors and the leg behaviors of the target user into the behavior fusion network to perform behavior fusion, so as to obtain the AI human behaviors.
The first behavior recognition model comprises a skeleton recognition network and a body recognition network;
the step of performing human body recognition on the user behavior to obtain human body data, and inputting the human body data into a first behavior recognition model to obtain model human body behavior, comprising the following steps:
human body recognition is carried out on the user behaviors to obtain skeleton data and trunk data;
inputting the bone data into the bone recognition network to obtain the human body movement characteristic data of the target user output by the bone recognition network;
inputting the trunk data into the trunk identification network to obtain trunk characteristic data of the target user output by the trunk identification network;
and fusing the human body motion characteristic data and the trunk characteristic data of the target user to obtain the model human body behavior of the target user.
The skeleton recognition network comprises a behavior feature extraction layer, a position feature extraction layer and a feature superposition layer, wherein the output ends of the behavior feature extraction layer and the position feature extraction layer are respectively connected with the input end of the feature superposition layer, the behavior feature extraction layer is used for extracting behavior feature data, the position feature extraction layer is used for extracting position feature data, and the feature superposition layer is used for fusing the behavior feature data and the position feature data;
inputting the bone data into the bone recognition network to obtain the human motion characteristic data of the target user output by the bone recognition network, wherein the human motion characteristic data comprises the following components:
inputting the bone data into the behavior feature extraction layer to perform behavior feature extraction to obtain behavior feature data of the target user;
inputting the bone data into the position feature extraction layer for position feature extraction to obtain position feature data of the target user;
and inputting the behavior characteristic data and the position characteristic data of the target user into the characteristic superposition layer for characteristic superposition fusion to obtain the human motion characteristic data of the target user.
The obtaining the log data of the user behavior includes:
acquiring target occurrence time of the user behavior, and acquiring a sample log sequence of the target user at the target occurrence time;
and splitting the sample log sequence into a preset number of sample subsequences, and determining all the sample subsequences as log data of the target user.
Inputting the log data into a second behavior recognition model to obtain log human behaviors, wherein the log human behaviors comprise:
inputting each sample subsequence into the second behavior recognition model to obtain each behavior attribute of the target user output by the second behavior recognition model;
and carrying out behavior fusion according to the spatial characteristics of each behavior attribute to obtain the log human behavior of the target user.
The behavior attributes comprise gesture attributes and motion trail attributes;
performing behavior fusion according to the spatial characteristics of each behavior attribute to obtain the log human behavior of the target user, including:
determining each gesture behavior of the target user according to the spatial characteristics of the gesture attribute in each behavior attribute, and determining each track behavior of the target user according to the spatial characteristics of the motion track attribute in each behavior attribute;
And performing behavior fusion on all the gesture behaviors of the target user, performing behavior fusion on all the track behaviors of the target user to obtain gesture track behaviors, and determining the gesture track behaviors as the log human body behaviors.
The monitoring and analyzing the user behavior of the target user according to the AI human behavior, the model human behavior and the log human behavior includes:
performing pairwise similarity comparison on the AI human body behaviors, the model human body behaviors and the log human body behaviors to obtain a first similarity comparison value, a second similarity comparison value and a third similarity comparison value;
if at least one difference value of the difference values of any two similarity comparison values among the first similarity comparison value, the second similarity comparison value and the third similarity comparison value is larger than a preset value, determining that abnormality exists in behavior monitoring;
if the difference value of any two similarity comparison values among the first similarity comparison value, the second similarity comparison value and the third similarity comparison value is smaller than or equal to the preset value, determining that no abnormality exists in behavior monitoring.
In a third aspect, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor implements the analysis method based on AI image recognition behavior monitoring according to the second aspect when the program is executed.
In a fourth aspect, the present invention also provides a computer-readable storage medium, which includes a computer program that, when executed by the processor, implements the analysis method based on AI image recognition behavior monitoring of the second aspect.
In a fifth aspect, the present invention also provides a computer program product, which comprises a computer program, which when executed by the processor implements the analysis method based on AI image recognition behavior monitoring of the second aspect.
The invention provides an analysis system based on AI image recognition behavior monitoring and a method thereof, wherein the system comprises an AI behavior recognition device, a human body behavior recognition device and a log behavior recognition device, the AI behavior recognition device is connected with the human body behavior recognition device, and the human body behavior recognition device is connected with the log behavior recognition device; the AI behavior recognition device comprises an AI image recognition device and an AI behavior recognition model; the human body behavior recognition device comprises a human body data recognition device and a first behavior recognition model; the log behavior recognition device comprises a log data acquisition device and a second behavior recognition model; the AI behavior recognition device is used for inputting the AI human body image obtained by the user behavior recognition of the target user by the AI image recognition device into the AI behavior recognition model to obtain AI human body behaviors; the human body behavior recognition device is used for inputting human body data obtained by recognizing the user behavior by the human body data recognition device into the first behavior recognition model to obtain model human body behaviors; the log behavior recognition device is used for inputting the log data acquired by the log data acquisition device according to the user behavior into the second behavior recognition model to acquire log human body behaviors; and monitoring and analyzing the user behaviors of the target user according to the AI human behaviors, the model human behaviors and the log human behaviors. The analysis system based on AI image recognition behavior monitoring can integrate AI human behaviors, model human behaviors and log human behaviors of a user into a whole for analysis, and eliminates deviation of behavior monitoring analysis, so that the behavior of the user can be accurately monitored and analyzed, and the accurate behavior monitoring analysis of the user behavior is realized.
Drawings
In order to more clearly illustrate the technical solutions of the present invention, the following description will be given with a brief introduction to the drawings used in the embodiments or the description of the prior art, it being obvious that the drawings in the following description are some embodiments of the present invention, and that other drawings can be obtained from these drawings without the inventive effort of a person skilled in the art.
FIG. 1 is a schematic diagram of an analysis system based on AI image recognition behavior monitoring provided by the invention;
FIG. 2 is a flow chart of an analysis method based on AI image recognition behavior monitoring provided by the invention;
fig. 3 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The embodiments of the present invention provide embodiments of an analysis system based on AI image recognition behavior monitoring, it being noted that although a logical sequence is shown in the flow chart, under certain data, the steps shown or described may be accomplished in a different order than that shown or described herein.
Referring to fig. 1, fig. 1 is a schematic structural diagram of an analysis system based on AI image recognition behavior monitoring provided by the present invention. The analysis system based on AI image recognition behavior monitoring provided by the embodiment of the invention comprises an artificial intelligence AI behavior recognition device 10, a human body behavior recognition device 20 and a log behavior recognition device 30. The AI behavior recognition device 10 is connected to the human behavior recognition device 20, and the human behavior recognition device 20 is connected to the log behavior recognition device 30.
In one embodiment, the AI behavior-recognition device 10 includes an AI image-recognition device 101 and an AI behavior-recognition model 102. The human behavior recognition device 20 includes a human data recognition device 201 and a first behavior recognition model 202. The log behavior recognition means 30 includes log data acquisition means 301 and a second behavior recognition model 302.
In one embodiment, the AI image recognition device 101 recognizes the user behavior of the target user, and obtains an AI human body image.
In one embodiment, the AI-human-body image obtained by the AI-behavior recognition device 10 AI-image recognition device 101 is input into the AI-behavior recognition model 102 to obtain AI-human-body behavior.
In an embodiment, the human body data recognition device 201 recognizes the user behavior of the target user, and obtains human body data.
In one embodiment, the human behavior recognition device 20 inputs the human data obtained by the human data recognition device 201 to the first behavior recognition model 202 to obtain the model human behavior.
In one embodiment, the log data acquisition means 301 acquires log data according to the user behavior of the target user.
In an embodiment, the log behavior recognition device 30 inputs the log data obtained by the log data obtaining device 301 to the second behavior recognition model 302, so as to obtain the log human body behavior.
In one embodiment, the analysis system based on AI image recognition behavior monitoring monitors and analyzes user behavior of the target user based on AI human behavior, model human behavior, and log human behavior.
The analysis system based on the AI image recognition behavior monitoring provided by the embodiment of the invention comprises an AI behavior recognition device, a human body behavior recognition device and a log behavior recognition device, wherein the AI behavior recognition device is connected with the human body behavior recognition device, and the human body behavior recognition device is connected with the log behavior recognition device; the AI behavior recognition device comprises an AI image recognition device and an AI behavior recognition model; the human body behavior recognition device comprises a human body data recognition device and a first behavior recognition model; the log behavior recognition device comprises a log data acquisition device and a second behavior recognition model; the AI behavior recognition device is used for inputting the AI human body image obtained by the user behavior recognition of the target user by the AI image recognition device into the AI behavior recognition model to obtain AI human body behaviors; the human body behavior recognition device is used for inputting human body data obtained by recognizing the user behavior by the human body data recognition device into the first behavior recognition model to obtain model human body behaviors; the log behavior recognition device is used for inputting the log data acquired by the log data acquisition device according to the user behavior into the second behavior recognition model to acquire log human body behaviors; and monitoring and analyzing the user behaviors of the target user according to the AI human behaviors, the model human behaviors and the log human behaviors. The analysis system based on AI image recognition behavior monitoring can integrate AI human behaviors, model human behaviors and log human behaviors of a user into a whole for analysis, and eliminates deviation of behavior monitoring analysis, so that the behavior of the user can be accurately monitored and analyzed, and the accurate behavior monitoring analysis of the user behavior is realized.
Further, the analysis method based on the AI image recognition behavior monitoring provided by the invention and the analysis system based on the AI image recognition behavior monitoring provided by the invention are mutually correspondingly referred.
Fig. 2 is a schematic flow chart of an analysis method based on AI image recognition behavior monitoring, where the analysis method based on AI image recognition behavior monitoring includes:
step 201, obtaining user behavior of a target user;
step 202, carrying out AI identification on the user behavior to obtain an AI human body image, and inputting the AI human body image into an AI behavior identification model to obtain an AI human body behavior;
step 203, performing human body recognition on the user behavior to obtain human body data, and inputting the human body data into a first behavior recognition model to obtain a model human body behavior;
step 204, obtaining log data of the user behaviors, and inputting the log data into a second behavior recognition model to obtain log human behaviors;
and 205, monitoring and analyzing the user behaviors of the target user according to the AI human behaviors, the model human behaviors and the log human behaviors.
The analysis system based on the AI image recognition behavior monitoring comprises an AI behavior recognition device, a human body behavior recognition device and a log behavior recognition device. The AI behavior recognition device is connected with the human behavior recognition device, and the human behavior recognition device is connected with the log behavior recognition device. The AI behavior recognition device includes an AI image recognition device and an AI behavior recognition model. The human behavior recognition device comprises a human body data recognition device and a first behavior recognition model. The log behavior recognition means includes log data acquisition means and a second behavior recognition model.
Further, when monitoring analysis is required to be performed on the user behavior of the user, the analysis system acquires the user behavior of the target user.
Further, the analysis system performs AI identification on the user behaviors to obtain AI human body images, and inputs the AI human body images into the AI behavior identification model to obtain AI human body behaviors.
Further, the analysis system performs human body recognition on the user behaviors to obtain human body data, and the human body data is input into the first behavior recognition model to obtain model human body behaviors.
Further, log data of the user behaviors are analyzed by the analysis system, and are input into the second behavior recognition model to obtain log human behaviors.
Further, the analysis system monitors and analyzes the user behaviors of the target user according to the combination of the AI human behaviors, the model human behaviors and the log human behaviors.
It should be noted that, the AI behavior recognition model is obtained by training an AI image sample and a corresponding behavior label thereof; the first behavior recognition model is obtained by training a human body data sample and a corresponding behavior label; the second behavior recognition model is obtained by training a log data sample and corresponding user behavior attribute sample data for a tag label.
The specific training process of the AI behavior recognition model is as follows:
shooting by a plurality of prearranged AI shooting devices with different directions, wherein the video acquisition frame rate is 30Hz, the original video size is 2304 multiplied by 1296 pixels, the original AI behavior video data can be obtained, and then the original AI behavior video can be marked frame by frame, so that an AI behavior image sample and a corresponding AI behavior label are obtained. Further, all the acquired sample data are divided into a test set, a training set and a verification set. In this embodiment, 6218 AI-behavior image samples may be taken as a training set, and the remaining 621 AI-behavior image samples are divided into a verification set, 612 AI-behavior image samples are taken as a test set, and the corresponding images and labeling information are placed under the corresponding folders.
And taking the AI behavior image samples and the corresponding AI behavior image samples as a group of training samples, namely taking each AI behavior image sample with the AI behavior label as a group of training samples, thereby obtaining a plurality of groups of training samples. In the embodiment of the invention, the AI behavior image samples are in one-to-one correspondence with the AI behavior tags carried by the AI behavior image samples.
Then, after multiple sets of training samples are obtained, multiple sets of training samples are sequentially input into the AI behavior recognition model, and the AI behavior recognition model is trained by utilizing the multiple sets of training samples, namely: the AI behavior image samples in each group of training samples and the AI behavior labels carried by the AI behavior image samples are simultaneously input into an AI behavior recognition model, model parameters in the AI behavior recognition model are adjusted by calculating loss function values according to each output result in the AI behavior recognition model, and under the condition that preset training termination conditions are met, the whole training process of the AI behavior recognition model is finally completed, and a trained AI behavior recognition model is obtained.
Based on the foregoing embodiment, as an alternative embodiment, training the AI behavior recognition model with multiple sets of training samples includes:
the preset loss function described in the embodiment of the invention refers to a loss function preset in an AI behavior recognition model and is used for model evaluation; the preset threshold refers to a threshold preset by the model, and is used for obtaining a minimum loss value and completing model training; the preset times refer to the preset maximum times of model iterative training.
After a plurality of groups of training samples are obtained, for any group of training samples, the AI behavior image samples in each group of training samples and the AI behavior labels carried by the AI behavior image samples are simultaneously input into an AI behavior recognition model, and the prediction probability corresponding to the training samples is output. On the basis, a loss value is calculated according to the prediction probability corresponding to the training sample and the AI behavior label corresponding to the training sample by using a preset loss function. Further, after the loss value is obtained by calculation, the training process ends. And then, based on the loss value, adjusting the model parameters of the AI behavior recognition model to update the weight parameters of each layer of the model in the AI behavior recognition model, and then, carrying out the next training, and repeatedly iterating the training of the model.
In the training process, if the training result of a certain group of training samples meets the preset training termination condition, if the loss value obtained by corresponding calculation is smaller than the preset threshold value, or the current iteration number reaches the preset number, the loss value of the model can be controlled within the convergence range, and the model training is ended. At this time, the obtained model parameters can be used as model parameters of the trained AI behavior recognition model, and the AI behavior recognition model training is completed, so that the trained AI behavior recognition model is obtained.
The training process of the first behavior recognition model and the second behavior recognition model is the same.
The analysis method based on AI image recognition behavior monitoring provided by the embodiment of the invention obtains the user behavior of the target user; performing AI identification on the user behaviors to obtain AI human body images, and inputting the AI human body images into an AI behavior identification model to obtain AI human body behaviors; human body recognition is carried out on the user behaviors to obtain human body data, and the human body data are input into a first behavior recognition model to obtain model human body behaviors; acquiring log data of user behaviors, and inputting the log data into a second behavior recognition model to obtain log human behaviors; and monitoring and analyzing the user behaviors of the target user according to the AI human behaviors, the model human behaviors and the log human behaviors. The analysis method based on the AI image recognition behavior monitoring can integrate the AI human behavior, the model human behavior and the log human behavior of the user into a whole for analysis, eliminates deviation of behavior monitoring analysis, and can accurately monitor and analyze the behavior of the user, thereby realizing accurate behavior monitoring analysis on the behavior of the user.
Further, the AI identification of the user behavior recorded in step 202 is performed to obtain an AI human body image, and the AI human body image is input into an AI behavior identification model to obtain an AI human body behavior, which includes:
performing AI identification on the user behaviors to obtain the behavior actions of the target user and AI human body images of human body areas;
the shoulder is used as a first segmentation characteristic line and the waist is used as a second segmentation characteristic line, the AI human body image is subjected to characteristic segmentation, and a head characteristic image, a body part characteristic image and a leg characteristic image of the target user are obtained;
inputting the head characteristic image into the head recognition network to obtain head characteristic data of the target user output by the head recognition network;
inputting the body part characteristic image into the body part recognition network to obtain body part characteristic data of the target user output by the body part recognition network;
inputting the leg characteristic image into the leg identification network to obtain leg characteristic data of the target user output by the leg identification network;
and obtaining the AI human behavior according to the head characteristic data, the body part characteristic data and the leg characteristic data of the target user.
The AI behavior recognition model includes a head recognition network, a body part recognition network, and a leg recognition network. The head recognition network is used for recognizing head characteristic data, the body part recognition network is used for recognizing body part characteristic data, and the leg recognition network is used for recognizing leg characteristic data.
Specifically, the analysis system performs AI identification on the user behavior to obtain the behavior action of the target user and an AI human body image of the human body area.
Further, the analysis system performs feature segmentation on the AI human body image by taking the shoulder as a first segmentation feature line and the waist as a second segmentation feature line to obtain a head feature image, a body part feature image and a leg feature image of the target user.
Further, the analysis system inputs the head characteristic image into the head recognition network to obtain head characteristic data of the target user output by the head recognition network. Further, the analysis system inputs the body part characteristic image into the body part recognition network to obtain body part characteristic data of the target user output by the body part recognition network. Further, the analysis system inputs the leg characteristic image into the leg identification network to obtain leg characteristic data of the target user output by the leg identification network. Further, the analysis system obtains AI human behaviors according to the head characteristic data, the body part characteristic data and the leg characteristic data of the target user.
According to the embodiment of the invention, the AI human body behaviors of the target user are obtained through the AI behavior recognition model, and the AI human body behaviors are provided for behavior monitoring analysis, so that the AI human body behaviors of the user, the model human body behaviors and the log human body behaviors can be combined into a whole for analysis, deviation of behavior monitoring analysis is eliminated, and therefore, the behavior of the user can be accurately monitored and analyzed, and the accurate behavior monitoring analysis of the user behavior is realized.
Further, obtaining the AI human behavior according to the head feature data, the body part feature data, and the leg feature data of the target user includes:
inputting the head characteristic data, the body part characteristic data and the leg characteristic data of the target user into the behavior recognition network to perform behavior recognition, so as to respectively obtain the head behavior, the body part behavior and the leg behavior of the target user;
and inputting the head behaviors, the body part behaviors and the leg behaviors of the target user into the behavior fusion network to perform behavior fusion, so as to obtain the AI human behaviors.
The AI behavior recognition model includes a behavior recognition network and a behavior fusion network. The behavior recognition network is used for recognizing head behaviors, body part behaviors and leg behaviors, and the behavior fusion network is used for fusing the head behaviors, the body part behaviors and the leg behaviors.
Specifically, the analysis system inputs the head characteristic data of the target user into the behavior recognition network to perform behavior recognition to obtain the head behavior of the target user, simultaneously inputs the body part characteristic data into the behavior recognition network to perform behavior recognition to obtain the body part behavior of the target user, and simultaneously inputs the leg characteristic data into the behavior recognition network to perform behavior recognition to obtain the leg behavior of the target user.
Further, the analysis system inputs the head behaviors, the body part behaviors and the leg behaviors of the target user into the behavior fusion network to conduct behavior fusion, and AI human behaviors are obtained.
According to the embodiment of the invention, the AI human body behaviors of the target user are obtained through the AI behavior recognition model, and the AI human body behaviors are provided for behavior monitoring analysis, so that the AI human body behaviors of the user, the model human body behaviors and the log human body behaviors can be combined into a whole for analysis, deviation of behavior monitoring analysis is eliminated, and therefore, the behavior of the user can be accurately monitored and analyzed, and the accurate behavior monitoring analysis of the user behavior is realized.
Further, the human body recognition of the user behavior recorded in step 203 is performed to obtain human body data, and the human body data is input into a first behavior recognition model to obtain a model human body behavior, which includes:
Human body recognition is carried out on the user behaviors to obtain skeleton data and trunk data;
inputting the bone data into the bone recognition network to obtain the human body movement characteristic data of the target user output by the bone recognition network;
inputting the trunk data into the trunk identification network to obtain trunk characteristic data of the target user output by the trunk identification network;
and fusing the human body motion characteristic data and the trunk characteristic data of the target user to obtain the model human body behavior of the target user.
It should be noted that the first behavior recognition model includes a bone recognition network and a body recognition network. The skeleton recognition network is used for recognizing human motion characteristic data, and the body recognition network is used for recognizing trunk characteristic data.
Specifically, the analysis system performs human body recognition on user behaviors to obtain skeleton data and trunk data of a target user.
Further, the analysis system inputs the bone data into the bone recognition network to obtain the human body movement characteristic data of the target user output by the bone recognition network. Further, the analysis system inputs the trunk data into the trunk identification network to obtain trunk feature data of the target user output by the trunk identification network. Further, the analysis system fuses the human motion characteristic data and the trunk characteristic data of the target user to obtain the model human behavior of the target user.
Further, inputting the bone data into the bone recognition network to obtain the human motion characteristic data of the target user output by the bone recognition network, including:
inputting the bone data into the behavior feature extraction layer to perform behavior feature extraction to obtain behavior feature data of the target user;
inputting the bone data into the position feature extraction layer for position feature extraction to obtain position feature data of the target user;
and inputting the behavior characteristic data and the position characteristic data of the target user into the characteristic superposition layer for characteristic superposition fusion to obtain the human motion characteristic data of the target user.
It should be noted that the bone recognition network includes a behavior feature extraction layer, a position feature extraction layer, and a feature superimposing layer, and output ends of the behavior feature extraction layer and the position feature extraction layer are respectively connected with input ends of the feature superimposing layer. The behavior feature extraction layer is used for extracting behavior feature data, the position feature extraction layer is used for extracting position feature data, and the feature superposition layer is used for fusing the behavior feature data and the position feature data.
Specifically, the analysis system inputs skeleton data into the behavior feature extraction layer to extract behavior features, so as to obtain behavior feature data of the target user. Further, the analysis system inputs the bone data into the position feature extraction layer to extract the position features, and position feature data of the target user is obtained. Further, the analysis system inputs the behavior feature data and the position feature data of the target user into the feature superposition layer for feature superposition fusion to obtain the human motion feature data of the target user.
According to the embodiment of the invention, the model human body behaviors of the target user are obtained through the first behavior recognition model, and the model human body behaviors are provided for behavior monitoring analysis, so that the AI human body behaviors, the model human body behaviors and the log human body behaviors of the user can be combined into a whole for analysis, deviation of the behavior monitoring analysis is eliminated, and therefore, the behavior of the user can be accurately monitored and analyzed, and the accurate behavior monitoring analysis of the user behavior is realized.
Further, the obtaining log data of the user behavior recorded in step 204 includes:
acquiring target occurrence time of the user behavior, and acquiring a sample log sequence of the target user at the target occurrence time;
and splitting the sample log sequence into a preset number of sample subsequences, and determining all the sample subsequences as log data of the target user.
Specifically, the analysis system obtains the target occurrence time of the user behavior, and obtains a sample log sequence of the target user at the target occurrence time. Further, the analysis system splits the sample log sequence into a preset number of sample subsequences, and determines all the sample subsequences as log data of the target user, wherein the preset number is set according to the actual situation.
Further, inputting the log data into the second behavior recognition model in step 204 to obtain a log human behavior, including:
inputting each sample subsequence into the second behavior recognition model to obtain each behavior attribute of the target user output by the second behavior recognition model;
and carrying out behavior fusion according to the spatial characteristics of each behavior attribute to obtain the log human behavior of the target user.
Specifically, the analysis system inputs each sample subsequence into the second behavior recognition model to obtain each behavior attribute of the target user output by the second behavior recognition model. Further, the analysis system performs behavior fusion according to the spatial characteristics of each behavior attribute to obtain the log human body behaviors of the target user.
According to the embodiment of the invention, the log human body behaviors of the target user are obtained through the second behavior recognition model, and the log human body behaviors are provided for behavior monitoring analysis, so that the AI human body behaviors of the user, the model human body behaviors and the log human body behaviors can be combined into a whole for analysis, deviation of the behavior monitoring analysis is eliminated, and therefore, the behavior of the user can be accurately monitored and analyzed, and the accurate behavior monitoring analysis of the user behavior is realized.
Further, performing behavior fusion according to the spatial feature of each behavior attribute to obtain the log human behavior of the target user, including:
determining each gesture behavior of the target user according to the spatial characteristics of the gesture attribute in each behavior attribute, and determining each track behavior of the target user according to the spatial characteristics of the motion track attribute in each behavior attribute;
and performing behavior fusion on all the gesture behaviors of the target user, performing behavior fusion on all the track behaviors of the target user to obtain gesture track behaviors, and determining the gesture track behaviors as the log human body behaviors.
It should be noted that the behavior attribute includes a gesture attribute and a motion trajectory attribute.
Specifically, the analysis system determines each gesture behavior of the target user according to the spatial characteristics of the gesture attribute in each behavior attribute, and determines each track behavior of the target user according to the spatial characteristics of the motion track attribute in each behavior attribute. Further, the analysis system performs behavior fusion on all gesture behaviors of the target user, and performs behavior fusion on all track behaviors of the target user to obtain gesture track behaviors. Further, the analysis system determines the gesture track behavior as a log human body behavior.
The embodiment of the invention provides the log human body behaviors for behavior monitoring analysis, so that the AI human body behaviors, the model human body behaviors and the log human body behaviors of the user can be combined into a whole for analysis, deviation of behavior monitoring analysis is eliminated, accurate behavior monitoring analysis can be carried out on the behaviors of the user, and accurate behavior monitoring analysis on the behaviors of the user is realized.
Further, the monitoring and analyzing the user behavior of the target user according to the AI human behavior, the model human behavior and the log human behavior in step 205 includes:
performing pairwise similarity comparison on the AI human body behaviors, the model human body behaviors and the log human body behaviors to obtain a first similarity comparison value, a second similarity comparison value and a third similarity comparison value;
if at least one difference value of the difference values of any two similarity comparison values among the first similarity comparison value, the second similarity comparison value and the third similarity comparison value is larger than a preset value, determining that abnormality exists in behavior monitoring;
if the difference value of any two similarity comparison values among the first similarity comparison value, the second similarity comparison value and the third similarity comparison value is smaller than or equal to the preset value, determining that no abnormality exists in behavior monitoring.
Specifically, the analysis system performs similarity comparison on the AI human body behaviors and the model human body behaviors to obtain a first similarity comparison value, performs similarity comparison on the AI human body behaviors and the log human body behaviors to obtain a second similarity comparison value, and performs similarity comparison on the model human body behaviors and the log human body behaviors to obtain a third similarity comparison value.
Further, the analysis system calculates a first difference between the first similarity comparison value and the second similarity comparison value, simultaneously calculates a second difference between the first similarity comparison value and the third similarity comparison value, and simultaneously calculates a third difference between the second similarity comparison value and the third similarity comparison value.
Further, if it is determined that the first difference is greater than a preset value, or/and the second difference is greater than the preset value, or/and the third difference is greater than the preset value, that is, at least one difference in the differences is greater than the preset value, the analysis system determines that the behavior monitoring is abnormal, that is, AI human behavior, or/and model human behavior, or/and log human behavior is inconsistent, where the preset value is set according to the actual setting.
Further, if it is determined that the first difference is smaller than or equal to the preset value, the second difference is smaller than or equal to the preset value, and the third difference is smaller than or equal to the preset value, the analysis system determines that the behavior monitoring is abnormal, that is, the AI human behavior, the model human behavior and the log human behavior are consistent.
According to the embodiment of the invention, the AI human body behaviors, the model human body behaviors and the log human body behaviors of the user are combined into a whole for analysis, so that deviation of behavior monitoring analysis is eliminated, the behaviors of the user can be accurately monitored and analyzed, and the accurate behavior monitoring analysis of the behaviors of the user is realized.
The specific embodiment of the analysis method based on the AI image recognition behavior monitoring provided by the invention is basically the same as the embodiments of the analysis system based on the AI image recognition behavior monitoring, and is not repeated herein.
Fig. 3 illustrates a physical schematic diagram of an electronic device, as shown in fig. 3, the electronic device may include: processor 310, communication interface (Communications Interface) 320, memory 330 and communication bus 340, wherein processor 310, communication interface 320, memory 330 accomplish communication with each other through communication bus 340. The processor 310 may invoke logic instructions in the memory 330 to perform an analysis system based on AI image recognition behavior monitoring, the method comprising:
acquiring user behaviors of a target user;
performing AI identification on the user behaviors to obtain AI human body images, and inputting the AI human body images into an AI behavior identification model to obtain AI human body behaviors;
Human body recognition is carried out on the user behaviors to obtain human body data, and the human body data are input into a first behavior recognition model to obtain model human body behaviors;
acquiring log data of the user behaviors, and inputting the log data into a second behavior recognition model to obtain log human behaviors;
monitoring and analyzing the user behaviors of the target user according to the AI human behaviors, the model human behaviors and the log human behaviors;
the AI behavior recognition model is obtained by training an AI image sample and a corresponding behavior label thereof; the first behavior recognition model is obtained by training a human body data sample and a corresponding behavior label thereof; the second behavior recognition model is obtained by training a log data sample and corresponding user behavior attribute sample data for a tag label.
Further, the logic instructions in the memory 330 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, randomAccessMemory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform the AI-image recognition behavior monitoring-based analysis system provided by the above methods, the method comprising:
acquiring user behaviors of a target user;
performing AI identification on the user behaviors to obtain AI human body images, and inputting the AI human body images into an AI behavior identification model to obtain AI human body behaviors;
human body recognition is carried out on the user behaviors to obtain human body data, and the human body data are input into a first behavior recognition model to obtain model human body behaviors;
acquiring log data of the user behaviors, and inputting the log data into a second behavior recognition model to obtain log human behaviors;
monitoring and analyzing the user behaviors of the target user according to the AI human behaviors, the model human behaviors and the log human behaviors;
the AI behavior recognition model is obtained by training an AI image sample and a corresponding behavior label thereof; the first behavior recognition model is obtained by training a human body data sample and a corresponding behavior label thereof; the second behavior recognition model is obtained by training a log data sample and corresponding user behavior attribute sample data for a tag label.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the above-provided AI-image-recognition-behavior-monitoring-based analysis system, the method comprising:
acquiring user behaviors of a target user;
performing AI identification on the user behaviors to obtain AI human body images, and inputting the AI human body images into an AI behavior identification model to obtain AI human body behaviors;
human body recognition is carried out on the user behaviors to obtain human body data, and the human body data are input into a first behavior recognition model to obtain model human body behaviors;
acquiring log data of the user behaviors, and inputting the log data into a second behavior recognition model to obtain log human behaviors;
monitoring and analyzing the user behaviors of the target user according to the AI human behaviors, the model human behaviors and the log human behaviors;
the AI behavior recognition model is obtained by training an AI image sample and a corresponding behavior label thereof; the first behavior recognition model is obtained by training a human body data sample and a corresponding behavior label thereof; the second behavior recognition model is obtained by training a log data sample and corresponding user behavior attribute sample data for a tag label.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. An analysis system based on AI image recognition behavior monitoring is characterized by comprising an artificial intelligent AI behavior recognition device, a human body behavior recognition device and a log behavior recognition device, wherein the AI behavior recognition device is connected with the human body behavior recognition device, and the human body behavior recognition device is connected with the log behavior recognition device;
the AI behavior recognition device comprises an AI image recognition device and an AI behavior recognition model; the human body behavior recognition device comprises a human body data recognition device and a first behavior recognition model; the log behavior recognition device comprises a log data acquisition device and a second behavior recognition model;
the AI behavior recognition device is used for inputting an AI human body image obtained by recognizing the user behavior of the target user by the AI image recognition device into the AI behavior recognition model to obtain AI human body behaviors;
the human body behavior recognition device is used for inputting human body data obtained by recognizing the user behavior by the human body data recognition device into the first behavior recognition model to obtain model human body behavior;
the log behavior recognition device is used for inputting the log data obtained by the log data obtaining device according to the user behavior into the second behavior recognition model to obtain log human behaviors;
The analysis system based on AI image recognition behavior monitoring is used for monitoring and analyzing the user behavior of the target user according to the AI human behavior, the model human behavior and the log human behavior.
2. An analysis method based on AI image recognition behavior monitoring, which is applied to the analysis system based on AI image recognition behavior monitoring as set forth in claim 1, comprising:
acquiring user behaviors of a target user;
performing AI identification on the user behaviors to obtain AI human body images, and inputting the AI human body images into an AI behavior identification model to obtain AI human body behaviors;
human body recognition is carried out on the user behaviors to obtain human body data, and the human body data are input into a first behavior recognition model to obtain model human body behaviors;
acquiring log data of the user behaviors, and inputting the log data into a second behavior recognition model to obtain log human behaviors;
monitoring and analyzing the user behaviors of the target user according to the AI human behaviors, the model human behaviors and the log human behaviors;
the AI behavior recognition model is obtained by training an AI image sample and a corresponding behavior label thereof; the first behavior recognition model is obtained by training a human body data sample and a corresponding behavior label thereof; the second behavior recognition model is obtained by training a log data sample and corresponding user behavior attribute sample data for a tag label.
3. The AI-image-recognition-behavior-monitoring-based analysis method of claim 2, wherein the AI-behavior-recognition model includes a head-recognition network, a body-part-recognition network, and a leg-recognition network;
the step of carrying out AI identification on the user behavior to obtain an AI human body image, and inputting the AI human body image into an AI behavior identification model to obtain an AI human body behavior, comprising the following steps:
performing AI identification on the user behaviors to obtain the behavior actions of the target user and AI human body images of human body areas;
the shoulder is used as a first segmentation characteristic line and the waist is used as a second segmentation characteristic line, the AI human body image is subjected to characteristic segmentation, and a head characteristic image, a body part characteristic image and a leg characteristic image of the target user are obtained;
inputting the head characteristic image into the head recognition network to obtain head characteristic data of the target user output by the head recognition network;
inputting the body part characteristic image into the body part recognition network to obtain body part characteristic data of the target user output by the body part recognition network;
inputting the leg characteristic image into the leg identification network to obtain leg characteristic data of the target user output by the leg identification network;
And obtaining the AI human behavior according to the head characteristic data, the body part characteristic data and the leg characteristic data of the target user.
4. The AI-image-recognition-behavior-monitoring-based analysis method of claim 3, wherein the AI behavior recognition model further includes a behavior recognition network and a behavior fusion network;
the obtaining the AI human behavior according to the head feature data, the body part feature data and the leg feature data of the target user includes:
inputting the head characteristic data, the body part characteristic data and the leg characteristic data of the target user into the behavior recognition network to perform behavior recognition, so as to respectively obtain the head behavior, the body part behavior and the leg behavior of the target user;
and inputting the head behaviors, the body part behaviors and the leg behaviors of the target user into the behavior fusion network to perform behavior fusion, so as to obtain the AI human behaviors.
5. The AI-image-recognition-behavior-monitoring-based analysis method of claim 2, wherein the first behavior recognition model includes a bone recognition network and a body recognition network;
the step of performing human body recognition on the user behavior to obtain human body data, and inputting the human body data into a first behavior recognition model to obtain model human body behavior, comprising the following steps:
Human body recognition is carried out on the user behaviors to obtain skeleton data and trunk data;
inputting the bone data into the bone recognition network to obtain the human body movement characteristic data of the target user output by the bone recognition network;
inputting the trunk data into the trunk identification network to obtain trunk characteristic data of the target user output by the trunk identification network;
and fusing the human body motion characteristic data and the trunk characteristic data of the target user to obtain the model human body behavior of the target user.
6. The analysis method based on AI image recognition behavior monitoring of claim 5, wherein the bone recognition network comprises a behavior feature extraction layer, a position feature extraction layer and a feature superposition layer, wherein the output ends of the behavior feature extraction layer and the position feature extraction layer are respectively connected with the input end of the feature superposition layer, the behavior feature extraction layer is used for extracting behavior feature data, the position feature extraction layer is used for extracting position feature data, and the feature superposition layer is used for fusing the behavior feature data and the position feature data;
inputting the bone data into the bone recognition network to obtain the human motion characteristic data of the target user output by the bone recognition network, wherein the human motion characteristic data comprises the following components:
Inputting the bone data into the behavior feature extraction layer to perform behavior feature extraction to obtain behavior feature data of the target user;
inputting the bone data into the position feature extraction layer for position feature extraction to obtain position feature data of the target user;
and inputting the behavior characteristic data and the position characteristic data of the target user into the characteristic superposition layer for characteristic superposition fusion to obtain the human motion characteristic data of the target user.
7. The AI-image-recognition-behavior-monitoring-based analysis method of claim 2, wherein the obtaining log data of the user behavior includes:
acquiring target occurrence time of the user behavior, and acquiring a sample log sequence of the target user at the target occurrence time;
and splitting the sample log sequence into a preset number of sample subsequences, and determining all the sample subsequences as log data of the target user.
8. The analysis method based on AI image recognition behavior monitoring of claim 7, wherein the inputting the log data into the second behavior recognition model to obtain the log human body behavior includes:
Inputting each sample subsequence into the second behavior recognition model to obtain each behavior attribute of the target user output by the second behavior recognition model;
and carrying out behavior fusion according to the spatial characteristics of each behavior attribute to obtain the log human behavior of the target user.
9. The AI-image-recognition-behavior-monitoring-based analysis method of claim 8, wherein the behavior attributes include a gesture attribute and a motion trajectory attribute;
performing behavior fusion according to the spatial characteristics of each behavior attribute to obtain the log human behavior of the target user, including:
determining each gesture behavior of the target user according to the spatial characteristics of the gesture attribute in each behavior attribute, and determining each track behavior of the target user according to the spatial characteristics of the motion track attribute in each behavior attribute;
and performing behavior fusion on all the gesture behaviors of the target user, performing behavior fusion on all the track behaviors of the target user to obtain gesture track behaviors, and determining the gesture track behaviors as the log human body behaviors.
10. The analysis method based on AI image recognition behavior monitoring according to any one of claims 2 to 9, wherein the monitoring analysis of the user behavior of the target user based on the AI human behavior, the model human behavior, and the log human behavior includes:
performing pairwise similarity comparison on the AI human body behaviors, the model human body behaviors and the log human body behaviors to obtain a first similarity comparison value, a second similarity comparison value and a third similarity comparison value;
if at least one difference value of the difference values of any two similarity comparison values among the first similarity comparison value, the second similarity comparison value and the third similarity comparison value is larger than a preset value, determining that abnormality exists in behavior monitoring;
if the difference value of any two similarity comparison values among the first similarity comparison value, the second similarity comparison value and the third similarity comparison value is smaller than or equal to the preset value, determining that no abnormality exists in behavior monitoring.
CN202310739120.2A 2023-06-20 2023-06-20 Analysis system and method based on AI image recognition behavior monitoring Active CN116740813B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310739120.2A CN116740813B (en) 2023-06-20 2023-06-20 Analysis system and method based on AI image recognition behavior monitoring

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310739120.2A CN116740813B (en) 2023-06-20 2023-06-20 Analysis system and method based on AI image recognition behavior monitoring

Publications (2)

Publication Number Publication Date
CN116740813A true CN116740813A (en) 2023-09-12
CN116740813B CN116740813B (en) 2024-01-05

Family

ID=87914773

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310739120.2A Active CN116740813B (en) 2023-06-20 2023-06-20 Analysis system and method based on AI image recognition behavior monitoring

Country Status (1)

Country Link
CN (1) CN116740813B (en)

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111860118A (en) * 2020-06-03 2020-10-30 安徽碧耕软件有限公司 Human behavior analysis method based on artificial intelligence
WO2021004510A1 (en) * 2019-07-09 2021-01-14 中山大学 Sensor-based separately deployed human body behavior recognition health management system
CN112488827A (en) * 2021-01-13 2021-03-12 润邦汇金金融服务外包(北京)有限公司 Artificial intelligence detects bank outlet wind accuse and management system
CN215068521U (en) * 2021-06-25 2021-12-07 深圳乐美尚科技有限公司 Intelligent security device based on AI face recognition/behavior analysis
CN114338248A (en) * 2022-03-15 2022-04-12 北京大学 User abnormal behavior detection method and device based on machine learning
US20220125359A1 (en) * 2019-12-02 2022-04-28 Eknauth Persaud Systems and methods for automated monitoring of human behavior
CN114511877A (en) * 2021-12-31 2022-05-17 特斯联科技集团有限公司 Behavior recognition method and device, storage medium and terminal
CN114724241A (en) * 2022-03-29 2022-07-08 平安科技（深圳）有限公司 Motion recognition method, device, equipment and storage medium based on skeleton point distance
WO2023273075A1 (en) * 2021-06-30 2023-01-05 深圳市商汤科技有限公司 Behavior recognition method and apparatus, and computer device and storage medium
CN116030370A (en) * 2021-10-26 2023-04-28 国网冀北电力有限公司计量中心 Behavior recognition method and device based on multi-target tracking and electronic equipment

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2021004510A1 (en) * 2019-07-09 2021-01-14 中山大学 Sensor-based separately deployed human body behavior recognition health management system
US20220125359A1 (en) * 2019-12-02 2022-04-28 Eknauth Persaud Systems and methods for automated monitoring of human behavior
CN111860118A (en) * 2020-06-03 2020-10-30 安徽碧耕软件有限公司 Human behavior analysis method based on artificial intelligence
CN112488827A (en) * 2021-01-13 2021-03-12 润邦汇金金融服务外包(北京)有限公司 Artificial intelligence detects bank outlet wind accuse and management system
CN215068521U (en) * 2021-06-25 2021-12-07 深圳乐美尚科技有限公司 Intelligent security device based on AI face recognition/behavior analysis
WO2023273075A1 (en) * 2021-06-30 2023-01-05 深圳市商汤科技有限公司 Behavior recognition method and apparatus, and computer device and storage medium
CN116030370A (en) * 2021-10-26 2023-04-28 国网冀北电力有限公司计量中心 Behavior recognition method and device based on multi-target tracking and electronic equipment
CN114511877A (en) * 2021-12-31 2022-05-17 特斯联科技集团有限公司 Behavior recognition method and device, storage medium and terminal
CN114338248A (en) * 2022-03-15 2022-04-12 北京大学 User abnormal behavior detection method and device based on machine learning
CN114724241A (en) * 2022-03-29 2022-07-08 平安科技（深圳）有限公司 Motion recognition method, device, equipment and storage medium based on skeleton point distance

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
SYLMARIE DÁVILA-MONTERO ET AL: "Review and Challenges of Technologies for Real-Time Human Behavior Monitoring", 《 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS》, vol. 15, no. 1, pages 2 - 28, XP011847040, DOI: 10.1109/TBCAS.2021.3060617 *
张蔚澜: "商场监控系统中的人体异常行为检测方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》, vol. 2023, no. 01, pages 138 - 1816 *

Also Published As

Publication number Publication date
CN116740813B (en) 2024-01-05

Similar Documents

Publication Publication Date Title
Nguyen et al. Weakly supervised action localization by sparse temporal pooling network
CN110070029B (en) Gait recognition method and device
CN108256433B (en) Motion attitude assessment method and system
Chen et al. Adversarial posenet: A structure-aware convolutional network for human pose estimation
CN110097037B (en) Intelligent monitoring method and device, storage medium and electronic equipment
Taherkhani et al. A deep face identification network enhanced by facial attributes prediction
CN109919977B (en) Video motion person tracking and identity recognition method based on time characteristics
CN108499107B (en) Control method and device for virtual role in virtual reality and storage medium
Hasan et al. Incremental activity modeling and recognition in streaming videos
CN111160275B (en) Pedestrian re-recognition model training method, device, computer equipment and storage medium
CN111310658B (en) Method and device for updating action pattern recognition model
CN108229262A (en) A kind of pornographic video detecting method and device
CN109033955A (en) A kind of face tracking method and system
CN111401105B (en) Video expression recognition method, device and equipment
CN111401192A (en) Model training method based on artificial intelligence and related device
CN111104897A (en) Training method and device for child face recognition model and storage medium
CN111402156A (en) Restoration method and device for smear image, storage medium and terminal equipment
CN109101858B (en) Action recognition method and device
CN114359787A (en) Target attribute identification method and device, computer equipment and storage medium
CN116740813B (en) Analysis system and method based on AI image recognition behavior monitoring
US20160140395A1 (en) Adaptive sampling for efficient analysis of ego-centric videos
CN116363712A (en) Palmprint palm vein recognition method based on modal informativity evaluation strategy
CN108197593B (en) Multi-size facial expression recognition method and device based on three-point positioning method
CN115424253A (en) License plate recognition method and device, electronic equipment and storage medium
KR20210018586A (en) Method and apparatus for identifying video content using biometric features of characters

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