CN112800814A - Method and device for identifying abnormal transportation behavior of package, terminal and storage medium - Google Patents

Method and device for identifying abnormal transportation behavior of package, terminal and storage medium Download PDF

Info

Publication number
CN112800814A
CN112800814A CN201911108587.7A CN201911108587A CN112800814A CN 112800814 A CN112800814 A CN 112800814A CN 201911108587 A CN201911108587 A CN 201911108587A CN 112800814 A CN112800814 A CN 112800814A
Authority
CN
China
Prior art keywords
behavior
data
model
abnormal transportation
acquiring
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.)
Pending
Application number
CN201911108587.7A
Other languages
Chinese (zh)
Inventor
邵耀辉
梁智
胡奉平
谭振辉
吴斯涵
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.)
SF Technology Co Ltd
SF Tech Co Ltd
Original Assignee
SF 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 SF Technology Co Ltd filed Critical SF Technology Co Ltd
Priority to CN201911108587.7A priority Critical patent/CN112800814A/en
Publication of CN112800814A publication Critical patent/CN112800814A/en
Pending legal-status Critical Current

Links

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
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • G06Q10/083—Shipping
    • 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

Abstract

The application discloses a method, a device, a terminal and a storage medium for identifying abnormal transportation behaviors of packages. The package abnormal transportation behavior identification method comprises the following steps: acquiring acceleration data of a parcel to be identified; inputting the acceleration data into a trained behavior recognition model, wherein the acceleration data is used for the behavior recognition model to recognize the motion acceleration of the parcel to be recognized in each direction, and the motion acceleration is used for the behavior recognition model to determine the abnormal transportation behavior of the parcel to be recognized; and acquiring the abnormal transportation behavior of the parcel to be identified, which is determined by the behavior identification model based on the acceleration data. The method and the system can identify the abnormal transportation behavior of the package, and provide convenience for accurately determining the responsible person or responsible object of package damage in the logistics industry.

Description

Method and device for identifying abnormal transportation behavior of package, terminal and storage medium
Technical Field
The embodiment of the application relates to the technical field of artificial intelligence, in particular to a method, a device, a terminal and a storage medium for identifying abnormal transportation behaviors of a package.
Background
With the continuous improvement of the living standard of people, online shopping becomes a new living mode, and the rapid development of online shopping drives the rapid development of the logistics industry. In order to meet the fine operation of the logistics industry, abnormal conditions in the transportation process of the packages are often detected, such as whether the packages are abnormally transported by being thrown by people, kicked by people, bumpy in a vehicle or sorted by equipment, so that responsible persons or responsible objects damaged by the packages can be determined conveniently.
However, in the prior art, the severity of the handling of the package by a person or device can only be determined by incorporating a sensor in the package to detect the acceleration values instantaneously received by the package. However, the simple acceleration detection cannot detect the abnormal transportation behavior of the parcel (such as being thrown by people, kicked by people, bumpy in the car or sorted by equipment), and cannot accurately determine the responsible person or responsible object of the damaged parcel.
Disclosure of Invention
The embodiment of the application provides a method, a device, a terminal and a storage medium for identifying abnormal transportation behaviors of a package, and aims to solve the problem that the abnormal transportation behaviors of the package cannot be identified in the prior art.
In order to achieve the above object, an embodiment of the present application provides a method for identifying an abnormal transportation behavior of a package, including:
acquiring acceleration data of a parcel to be identified;
inputting the acceleration data into a trained behavior recognition model, wherein the acceleration data is used for the behavior recognition model to recognize the motion acceleration of the parcel to be recognized in each direction, and the motion acceleration is used for the behavior recognition model to determine the abnormal transportation behavior of the parcel to be recognized;
and acquiring the abnormal transportation behavior of the parcel to be identified, which is determined by the behavior identification model based on the acceleration data.
In some embodiments of the present application, the inputting the acceleration data into the trained behavior recognition model further comprises:
acquiring training data of a model to be trained;
and inputting the training data into the model to be trained for learning to obtain the behavior recognition model.
In some embodiments of the present application, the obtaining training data of the model to be trained includes:
acquiring preset specific abnormal transportation behaviors and behavior segments of the specific abnormal transportation behaviors, and acquiring preset sampling duration and sampling rate of the behavior segments;
acquiring segment acceleration data of the behavior segment according to the sampling duration and the sampling rate;
and integrating the fragment acceleration data of all the behavior fragments of the specific abnormal transportation behavior into high-dimensional data serving as training data of a model to be trained.
In some embodiments of the present application, the inputting the training data into the model to be trained for learning to obtain the behavior recognition model includes:
inputting the training data into a feature characterization submodel of the model to be trained;
acquiring target characteristics output by the characteristic characterization submodel based on the training data;
and inputting the target characteristics into an information memory sub-model of the model to be trained for learning to obtain the behavior recognition model.
In some embodiments of the present application, the obtaining of the target feature output by the feature characterization submodel based on the training data includes:
acquiring original features of the training data and extracted features of the training data through the feature characterization submodel;
and performing feature superposition on the original features and the extracted features through the feature characterization submodel to obtain the target features.
In some embodiments of the present application, the obtaining the extracted features of the training data includes:
acquiring a first extraction weight of the training data according to a preset first weight range;
acquiring first feature data from the training data according to the first extraction weight;
acquiring second characteristic data from the first characteristic data through a characteristic extraction layer of the characteristic characterization submodel;
acquiring a second extraction weight of the second feature data according to a preset second weight range;
and acquiring data from the second feature data according to the second extraction weight to serve as the extraction feature of the training data.
In some embodiments of the present application, the obtaining of the original features of the training data includes:
acquiring a third extraction weight of the training data according to a preset third weight range;
and acquiring data from the training data according to the third extraction weight to serve as the original features of the training data.
In addition, in order to achieve the above object, an embodiment of the present application further provides an abnormal transportation behavior recognition apparatus for a package, where the abnormal transportation behavior recognition apparatus for a package includes:
the identification data acquisition unit is used for acquiring acceleration data of the parcel to be identified;
the input unit is used for inputting the acceleration data into a trained behavior recognition model, the acceleration data is used for the behavior recognition model to recognize the motion acceleration of the parcel to be recognized in each direction, and the motion acceleration is used for the behavior recognition model to determine the abnormal transportation behavior of the parcel to be recognized;
and the identification unit is used for acquiring the abnormal transportation behavior of the parcel to be identified, which is determined by the behavior identification model based on the acceleration data.
In some embodiments of the present application, the apparatus for identifying abnormal transportation behavior of a package further includes:
the training data acquisition unit is used for acquiring training data of the model to be trained;
and the model training unit is used for inputting the training data into the model to be trained for learning to obtain the behavior recognition model.
In some embodiments of the present application, the training data obtaining unit is specifically configured to:
acquiring preset specific abnormal transportation behaviors and behavior segments of the specific abnormal transportation behaviors, and acquiring preset sampling duration and sampling rate of the behavior segments;
acquiring segment acceleration data of the behavior segment according to the sampling duration and the sampling rate;
and integrating the fragment acceleration data of all the behavior fragments of the specific abnormal transportation behavior into high-dimensional data serving as training data of a model to be trained.
In some embodiments of the present application, the model training unit is specifically configured to:
inputting the training data into a feature characterization submodel of the model to be trained;
acquiring target characteristics output by the characteristic characterization submodel based on the training data;
and inputting the target characteristics into an information memory sub-model of the model to be trained for learning to obtain the behavior recognition model.
In some embodiments of the present application, the model training unit is specifically configured to:
acquiring original features of the training data and extracted features of the training data through the feature characterization submodel;
and performing feature superposition on the original features and the extracted features through the feature characterization submodel to obtain the target features.
In some embodiments of the present application, the model training unit is specifically configured to:
acquiring a first extraction weight of the training data according to a preset first weight range;
acquiring first feature data from the training data according to the first extraction weight;
acquiring second characteristic data from the first characteristic data through a characteristic extraction layer of the characteristic characterization submodel;
acquiring a second extraction weight of the second feature data according to a preset second weight range;
and acquiring data from the second feature data according to the second extraction weight to serve as the extraction feature of the training data.
In some embodiments of the present application, the model training unit is specifically configured to:
acquiring a third extraction weight of the training data according to a preset third weight range;
and acquiring data from the training data according to the third extraction weight to serve as the original features of the training data.
In addition, in order to achieve the above object, an embodiment of the present application further provides an abnormal transportation behavior identification terminal for a package, where the abnormal transportation behavior identification terminal for a package includes: the package abnormal transportation behavior identification program realizes the steps of the package abnormal transportation behavior identification method when being executed by the processor.
In addition, to achieve the above object, an embodiment of the present application further provides a storage medium, where the storage medium stores an abnormal transportation behavior recognition program of a package, and the abnormal transportation behavior recognition program of the package, when executed by a processor, implements the steps of the abnormal transportation behavior recognition method of the package as described above.
The method comprises the steps of acquiring acceleration data of a parcel to be identified, and inputting the acceleration data of the parcel to be identified into a trained behavior identification model; analyzing the acquired acceleration data of the parcel to be recognized based on the trained behavior recognition model, and recognizing the motion acceleration of the parcel to be recognized in each direction; according to the motion acceleration of the parcel to be identified in each direction, identifying the abnormal transportation behavior of the parcel to be identified, namely determining the condition of the parcel to be identified, which is operated by people or equipment; therefore, the responsible person or responsible object of the package damage can be accurately determined according to the abnormal transportation behavior of the package to be identified, and convenience is provided for logistics industry management.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments of the present application will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a schematic flow chart illustrating an embodiment of a method for identifying abnormal transportation behavior of a package according to the present application;
FIG. 2 is a schematic diagram of an embodiment of a network structure of a model to be trained according to the present application;
FIG. 3 is a schematic flowchart of an embodiment of training a model to be trained in the package abnormal transportation behavior recognition method according to the present application;
FIG. 4 is a flowchart illustrating an embodiment of step 202 in the present application;
FIG. 5 is a schematic view of an embodiment of an abnormal transportation behavior recognition device for packages according to the present application;
fig. 6 is a schematic structural diagram of an embodiment of a terminal provided in the embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
In the description of the embodiments of the present application, it is to be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", and the like, indicate orientations and positional relationships based on those shown in the drawings, and are used merely for convenience of description and for simplicity of description, and do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore should not be considered as limiting the present application. Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first", "second", may explicitly or implicitly include one or more of the described features. In the description of the embodiments of the present application, "a plurality" means two or more unless specifically defined otherwise.
In the embodiments of the present application, the word "exemplary" is used to mean "serving as an example, instance, or illustration. Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments. The following description is presented to enable any person skilled in the art to make and use the application. In the following description, details are set forth for the purpose of explanation. It will be apparent to one of ordinary skill in the art that the present application may be practiced without these specific details. In other instances, well-known structures and processes are not shown in detail to avoid obscuring the description of the embodiments of the present application with unnecessary detail. Thus, the present embodiments are not intended to be limited to the embodiments shown, but are to be accorded the widest scope consistent with the principles and features disclosed in the embodiments herein.
The embodiments of the present application provide a method, an apparatus, a terminal and a storage medium for identifying an abnormal transportation behavior of a package, which are described in detail below.
First, an embodiment of the present application provides a method for identifying an abnormal transportation behavior of a package, and referring to fig. 1, fig. 1 is a schematic flow diagram of an embodiment of the method for identifying an abnormal transportation behavior of a package according to the present application.
While a logical order is shown in the flow chart, in some cases, the steps shown or described may be performed in a different order than presented herein.
The method for identifying the abnormal transportation behavior of the package is applied to a terminal or a server, and the terminal may include a mobile terminal such as a mobile phone, a tablet computer, a notebook computer, a palm computer, a Personal Digital Assistant (PDA), and the like, and a fixed terminal such as a Digital TV, a desktop computer, and the like. In the embodiments of the method for identifying the abnormal transportation behavior of the package, for convenience of description, the embodiments are described with an abnormal transportation behavior identification terminal (hereinafter referred to as a terminal) of the package as an execution subject. In an embodiment of the method for identifying abnormal transportation behavior of a package, the method for identifying abnormal transportation behavior of a package includes:
and 101, acquiring acceleration data of the parcel to be identified.
The acceleration data refers to the instantaneous acceleration of the parcel to be identified on the x axis, the instantaneous acceleration of the parcel to be identified on the y axis, the instantaneous acceleration of the parcel to be identified on the z axis and/or the resultant acceleration of the parcel to be identified (the resultant acceleration of the instantaneous acceleration on the x axis, the instantaneous acceleration on the y axis and the instantaneous acceleration on the z axis) at each moment of the manual or equipment operation process of the parcel to be identified during the transportation process of the parcel to be identified. The specific type of acceleration data to be collected can be determined according to the data required to be input by the trained behavior recognition model.
Specifically, there are various ways to acquire the acceleration data of the parcel to be identified, for example, a sensor is placed in the parcel to be identified, and the acceleration data of the parcel to be identified is collected by the sensor. In another example, a video of the process of operating the parcel to be identified is shot, and the acceleration data of the parcel to be identified is analyzed through the video.
102, inputting the acceleration data into a trained behavior recognition model, wherein the acceleration data is used for the behavior recognition model to recognize the motion acceleration of the parcel to be recognized in each direction, and the motion acceleration is used for the behavior recognition model to determine the abnormal transportation behavior of the parcel to be recognized.
Specifically, the acquired acceleration data of the parcel to be identified is input into the trained behavior identification model, so that the behavior identification model analyzes the motion acceleration of the parcel to be identified in each direction based on the acceleration data of the parcel to be identified, for example, at t1At the moment, the instantaneous acceleration of the parcel to be identified in the x-axis direction, the instantaneous acceleration in the y-axis direction and the instantaneous acceleration in the z-axis direction; at t2At the moment, the instantaneous acceleration of the parcel to be identified in the x-axis direction, the instantaneous acceleration in the y-axis direction and the instantaneous acceleration in the z-axis direction; at tnAnd at the moment, the instantaneous acceleration of the parcel to be identified in the x-axis direction, the instantaneous acceleration in the y-axis direction and the instantaneous acceleration in the z-axis direction. And enabling the behavior recognition model to determine the abnormal transportation behavior of the parcel to be recognized according to the relation between the motion acceleration of the parcel to be recognized in each direction and each preset abnormal transportation behavior. Before the trained behavior recognition model is adopted to recognize the abnormal transportation behavior of the package to be recognized, training samples need to be collected to train the model until the model converges, and the trained behavior recognition model which can be used for recognizing the abnormal transportation behavior of the package is obtained. The specific training process of the behavior recognition model may refer to the following embodiments of step 201 and step 202. Further, the behavior recognition model can analyze the abnormal transportation behavior of the parcel to be recognized based on the acceleration data of the parcel to be recognized, and simultaneously can detect the abnormal grade of the abnormal transportation behavior of the parcel to be recognized, such as: the abnormal transportation behavior of the package comprises artificial kicking, the abnormal grade of the artificial kicking is divided into high grade, medium grade and low grade, and a behavior identification modelThe type can identify that the abnormal transportation behavior of the package is artificial kicking, and the abnormal grade of the artificial kicking is high.
The behavior recognition model is a model for recognizing abnormal transportation behaviors of the packages according to the acquired acceleration speed; for example, according to the instantaneous acceleration of the parcel in the x axis, the y axis and the z axis respectively, the abnormal transportation behavior of the parcel is identified as follows: is vibrated by the belt conveyor, jolted by the inside of the vehicle, thrown by people, kicked by people, and slides down by a slope chute or a spiral chute. The abnormal transportation behavior refers to the operation condition of the parcel, such as being shaken by a belt conveyor, bumped in a car, thrown by a person, kicked by a person, slid by a slope chute and slid by a spiral chute, which is manually or mechanically operated during the transportation of the parcel.
103, acquiring abnormal transportation behaviors of the parcel to be identified, which are determined by the behavior identification model based on the acceleration data.
Referring to fig. 2, in this embodiment, fig. 2 shows a trained recognition model, and specifically, as an implementation manner, acceleration data of a package to be recognized is input into the trained behavior recognition model, after the acceleration data is subjected to feature characterization by a feature characterization sub-model of the trained behavior recognition model, an information memory sub-model of the trained behavior recognition model recognizes an abnormal transportation behavior of the package to be recognized, and outputs the abnormal transportation behavior of the package to be recognized.
In the embodiment of the application, acceleration data of the parcel to be identified is collected and input into a trained behavior identification model; analyzing the acquired acceleration data of the parcel to be recognized based on the trained behavior recognition model, and recognizing the motion acceleration of the parcel to be recognized in each direction; according to the motion acceleration of the parcel to be identified in each direction, identifying the abnormal transportation behavior of the parcel to be identified, namely determining the condition of the parcel to be identified, which is operated by people or equipment; therefore, the responsible person or responsible object of the package damage can be accurately determined according to the abnormal transportation behavior of the package to be identified, and convenience is provided for logistics industry management.
Further, in order that the trained behavior recognition model can accurately recognize the abnormal transportation behavior of the package to be recognized, before the abnormal transportation behavior is recognized by the trained behavior recognition model, training samples need to be collected to train the model until the model converges, and the trained behavior recognition model which can be used for recognizing the abnormal transportation behavior is obtained. Referring to fig. 3, fig. 3 is a schematic flow chart of an embodiment of training a model to be trained in the package abnormal transportation behavior method of the present application, where the step of training the model to be trained includes:
and 201, acquiring training data of a model to be trained.
The model to be trained refers to an untrained behavior recognition model. After the untrained behavior recognition model (i.e., the model to be trained) is trained, a model (i.e., the trained behavior recognition model) which can be used for recognizing the abnormal transportation behavior of the package is formed.
For convenience of description and distinction, in the following description, if it is not specifically indicated whether the behavior recognition model is a trained behavior recognition model or an untrained behavior recognition model, the behavior recognition model refers to the trained behavior recognition model, and the untrained behavior recognition model is referred to as a model to be trained.
In order to enable the trained behavior recognition model to distinguish various types of abnormal transportation behaviors, package acceleration data (such as the instantaneous acceleration of a package on an x-axis, the instantaneous acceleration on a y-axis and/or the instantaneous acceleration on a z-axis) of each abnormal transportation behavior is collected when the model to be trained is trained, and a corresponding abnormal transportation behavior label is attached to the package acceleration data of each abnormal transportation behavior. And taking the acquired parcel acceleration data as training data of the model to be trained, and forming a training data set of the model to be trained by the acquired parcel acceleration data of the abnormal transportation behaviors.
For ease of understanding, a specific embodiment is described. For example, by embedding a sensor in a parcel, parcel acceleration data for each type of abnormal transportation behavior (instantaneous acceleration of the parcel in the x, y, and z axes, respectively) is collected for the following six types of abnormal transportation behavior of the parcel:
(a) is vibrated by the belt conveyor: and placing the packages on different types of belt conveyors, and collecting package acceleration data during vibration.
(b) Bumping inside the vehicle: parcels are placed in carriages of different sizes and types, and as the vehicle travels, parcel acceleration data at bumps is collected.
(c) Is thrown by people: and throwing the packages at different throwing heights, postures and forces, and collecting acceleration data of the packages during throwing.
(d) Is kicked by people: selecting different ground surfaces, kicking the packages with different force and angle, and collecting the acceleration data of the packages when the package is kicked.
(e) Is slid by the slope sliding groove: in the transfer yard, select the slope spout of different height, angle, smoothness, do the parcel landing experiment, the parcel acceleration data when collecting the slope spout landing.
(f) Is slid by the spiral chute: in the transfer yard, spiral chutes with different heights, radians and smoothness are selected to perform a package sliding experiment, and package acceleration data when the spiral chutes slide is collected.
Training data refers to package acceleration data for each abnormal transportation behavior (e.g., instantaneous acceleration of the package in the x-axis, instantaneous acceleration in the y-axis, and/or instantaneous acceleration in the z-axis), and an abnormal transportation behavior signature for each package acceleration data.
Specifically, in some embodiments of the present application, the step of obtaining training data of the model to be trained may include:
a1, acquiring preset specific abnormal transportation behaviors and behavior segments of the specific abnormal transportation behaviors, and acquiring preset sampling duration and sampling rate of the behavior segments.
Due to the fact that the abnormal transportation behaviors are time-sequenced and have strong fragmentality in the embodiment of the application, in order to collect package acceleration data with higher reference value, specific sampling duration and sampling rate can be set according to specific conditions, and the sampling duration and the sampling rate can be set according to the specific conditions.
Specifically, according to the actual conditions encountered in the package transportation process, the manual or equipment operation conditions of the package are divided to obtain a plurality of specific abnormal transportation behaviors (such as being vibrated by a belt conveyor, being jolted in a car, being thrown manually, being kicked manually, being slipped down by a slope chute, being slipped down by a spiral chute and the like), and package acceleration data of each specific abnormal transportation behavior is collected to be further used as training data, so that the trained behavior recognition model has the capability of distinguishing the specific abnormal transportation behaviors.
Since abnormal transportation behavior is time-sequenced and has stronger fragmentation in the embodiment of the application, in order to collect more reference value parcel acceleration data, firstly collecting the behavior data of parcels in a specific abnormal transportation behavior period (including the instantaneous acceleration of parcels in an x-axis, the instantaneous acceleration in a y-axis, the instantaneous acceleration in a z-axis and/or the combined acceleration of parcels at various time points in the period can be set according to specific requirements), and then: and segmenting the behavior data of the package in each specific abnormal transportation behavior period to obtain a plurality of behavior segments of each specific abnormal transportation behavior, and presetting the sampling time length and the sampling rate of each behavior segment. The behavior data of each specific abnormal transportation behavior is segmented, and the existing sliding window slicing technology can be adopted to segment the behavior data of the package in each specific abnormal transportation behavior period into m behavior segments.
The specific abnormal transportation behavior refers to the abnormal transportation behavior of the package determined by dividing the manual or equipment operation condition of the package according to the actual condition encountered in the package transportation process; the model to be trained can learn the specific abnormal transportation behavior of the package.
The behavior data of the specific abnormal transportation behavior refers to a set of acceleration data of the package at each moment in the process of the specific abnormal transportation behavior; the acceleration data may include, among other things, the instantaneous acceleration in the x-axis, the instantaneous acceleration in the y-axis of the parcel to be identified, the instantaneous acceleration in the z-axis of the parcel to be identified, and/or the resultant acceleration of the parcel to be identified (the instantaneous acceleration in the x-axis, the instantaneous acceleration in the y-axis, and the resultant acceleration in the z-axis).
The behavior segment refers to each data segment obtained after the package is segmented according to the behavior data in the specific abnormal transportation behavior period. For example, behavior data with a duration of 10 seconds is segmented to obtain 5 data segments, the duration of each data segment is 2 seconds, and each 2-second data segment is defined as a behavior segment.
A2, acquiring segment acceleration data of the behavior segment according to the sampling duration and the sampling rate.
Acquiring acceleration data (including instantaneous acceleration of the parcel on an x axis, instantaneous acceleration of the parcel on a y axis, instantaneous acceleration of the parcel on a z axis and/or combined acceleration of the parcel, which can be set according to specific requirements) of the parcel in each behavior segment according to a preset sampling rate and sampling duration of each behavior segment, and thus obtaining the segment acceleration data of each behavior segment of the specific abnormal transportation behavior.
The piece acceleration data refers to acceleration data of the parcel collected under each action piece (instantaneous acceleration of the parcel on the x axis, instantaneous acceleration of the parcel on the y axis, instantaneous acceleration of the parcel on the z axis, and/or combined acceleration of the parcel, which can be set according to specific requirements).
For example, if the specific abnormal transportation behavior is an artificial throw, the process of the artificial throw is divided into 10 behavior segments of continuous movement behaviors, the preset sampling time of each behavior segment is 3 seconds, the preset sampling rate is 100 acceleration data collected per second, and 3 × 100 to 300 acceleration data are collected in each behavior segment. The total number of 10 behavior segments of a specific abnormal transportation behavior is 10 × 300, 3000 acceleration data. Each acceleration data, in turn, may include instantaneous acceleration wrapped in the x-axis, instantaneous acceleration in the y-axis, instantaneous acceleration in the z-axis, and a resultant acceleration (in the x-axis, y-axis, and z-axis).
A3, integrating the segment acceleration data of all the behavior segments of the specific abnormal transportation behavior into high-dimensional data to be used as training data of a model to be trained.
Specifically, the segment acceleration data of all behavior segments belonging to the same specific abnormal transportation behavior are integrated to form high-dimensional data, and each high-dimensional data is used as training data of the model to be trained. In the embodiment of the application, the segment acceleration data of all behavior segments belonging to the same specific abnormal transportation behavior are converted into image data which can be understood by a computer through the existing matrix transformation technology. Then, after integrating the segment acceleration data of all behavior segments of the same specific abnormal transportation behavior, the dimensionality of the formed high-dimensional data is as follows:
m*n*width*channel
wherein n is the image height, width is the image width, and channel is the image channel number; m represents the number of behavioral segments of the specific abnormal transportation behavior, and n × width represents segment acceleration data of one behavioral segment. Further, the "number of image channels" (channel) is determined according to the number of types of acceleration data acquired at each time.
Namely: the data dimension of the training data that finally enters the model to be trained is m n width channel.
For ease of understanding, the description is continued following the example of the above step S42. For example, if the specific abnormal transportation behavior has 10 behavior segments, m is 10. One behavior segment collected 3 × 100 — 300 acceleration data, and n × width 300. If the acceleration data collected at each moment includes: the instantaneous acceleration of the parcel in the x-axis, the instantaneous acceleration of y-axis, the instantaneous acceleration of z-axis, totally 3 types, then channel is 3; if the acceleration data collected at each point includes: the instantaneous acceleration of the envelope on the x-axis, the instantaneous acceleration on the y-axis, the instantaneous acceleration on the z-axis, the combined acceleration (on the x-axis, the y-axis and the z-axis), 4 types, the channel is 4.
In the embodiment of the application, the collected data are integrated into high-dimensional data, so that the model to be trained can better understand the training data, the model to be trained can be accurately trained subsequently, and a behavior recognition model is obtained.
202, inputting the training data into the model to be trained for learning, and obtaining the behavior recognition model.
Specifically, for ease of understanding, referring to fig. 2, fig. 2 is a schematic diagram of an embodiment of a network structure of a model to be trained according to the present application. The model to be trained comprises a characteristic characterization submodel and an information memory submodel, wherein X represents input data (namely training data), and Y represents output data (namely an identification result of abnormal transportation behaviors). After X (namely training data) is input into the model to be trained, firstly, data characterization is carried out on the feature characterization sub-model to obtain target features. And the target characteristics enter an information memory sub-model to be analyzed, and the information memory sub-model determines the relationship between the target characteristics and each abnormal transportation behavior according to the target characteristics (namely the information memory sub-model learns the relationship between the target characteristics and the abnormal transportation behaviors). And obtaining a model which is trained after the information memory sub-model is studied based on the training data set.
And obtaining the model which is trained after the model to be trained is studied based on the training data set. Because of each model parameter, based on the training data set, a model that completes the training is obtained. In order to improve the accuracy of the behavior recognition model for recognizing the abnormal transportation behavior, a model with the highest recognition rate for recognizing the abnormal transportation behavior of the package needs to be obtained from a plurality of trained models and used as the behavior recognition model. Specifically, verification sample data is acquired, input to the trained model (for the trained model to analyze based on the verification sample data and identify the abnormal transportation behavior of the package), and the identification rate of the trained model is checked. And then, selecting a model with the highest recognition rate for recognizing the abnormal transportation behavior of the package from the plurality of trained models as a behavior recognition model.
Wherein, step 201 and step 202 may be both executed before step 101, or step 201 and step 202 may be both executed after step 101 and before step 102, or step 201 may be executed before step 101 and step 202 may be executed after step 101 and before step 102.
In the embodiment of the application, before the behavior recognition model is used for recognizing the abnormal transportation behavior of the package to be recognized, acceleration data of the package in various abnormal transportation behaviors is collected in advance to serve as sample data, so that the model to be trained learns each abnormal transportation behavior aiming at the sample data to obtain the behavior recognition model for recognizing the abnormal transportation behavior of the package, and the behavior recognition model is ensured to be capable of accurately recognizing the abnormal transportation behavior of the package to be recognized.
In some embodiments of the application, as shown in fig. 4, the step of inputting the training data into the model to be trained for learning to obtain the behavior recognition model may further include:
401, inputting the training data into the feature characterization submodel of the model to be trained.
In order to enable the model to automatically screen out effective feature data for learning, as shown in fig. 2, in this embodiment, the feature characterization submodel is added to the model to be trained, so as to screen out training data.
Specifically, the training data obtained in step 201 is input into the model to be trained, so that the model to be trained learns based on the training data, and a behavior recognition model capable of recognizing the abnormal transportation behavior of the package is obtained. In order to improve the effectiveness of the data, the training data obtained in step a3 is input into the feature characterization submodel of the model to be trained.
And 402, acquiring target characteristics output by the characteristic characterization submodel based on the training data.
The target characteristics refer to characteristic data obtained after the training data are characterized by a characteristic characterization submodel of the model to be trained.
The feature characterization submodel forms target features through feature extraction, feature superposition and the like based on the training data input in step 401, namely the feature characterization submodel performs data characterization based on the training data to obtain the target features, and the target features are used as features to be learned by the model to be trained. Referring to fig. 2, the feature characterization submodel further includes a feature extraction layer and a feature overlay ([ ] indicates a feature overlay in fig. 2). The feature extraction layer is used for extracting data from X (i.e., training data), and the main structure of the feature extraction layer may be a Convolutional Neural Network (CNN), or another network that can implement feature extraction similar to a Convolutional Neural network. The process of obtaining the target feature by performing data characterization on the feature characterization submodel by using X (namely training data) is as follows:
(1) and X (namely training data) passes through the feature extraction layer to obtain extracted features.
(2) And (3) performing feature superposition on the X (namely training data) and the extracted features obtained in the step (1) to obtain target features.
In some embodiments of the present application, the step of obtaining the target feature output by the feature characterization submodel based on the training data may include:
and B1, acquiring the original features of the training data and the extracted features of the training data through the feature characterization submodel.
After the training data are input into the feature characterization submodel, on one hand, the training data (specifically, the data with a certain weight obtained from the training data) are subjected to feature extraction through a feature extraction layer, and then the data with a certain weight is obtained from the data extracted by the feature extraction layer and is used as the extraction features of the training data; on the other hand, certain weight data is obtained from the training data as the original features of the training data.
The original features refer to data with a certain weight obtained from training data.
Extracting features, namely extracting data from training data (specifically, data with certain weight obtained from the training data) through a feature extraction layer; and acquiring data with certain weight from the data extracted by the feature extraction layer.
Specifically, the step of obtaining the extracted features of the training data may include:
acquiring a first extraction weight of the training data according to a preset first weight range; acquiring first feature data from the training data according to the first extraction weight; acquiring second characteristic data from the first characteristic data through a characteristic extraction layer of the characteristic characterization submodel; acquiring a second extraction weight of the second feature data according to a preset second weight range; and acquiring data from the second feature data according to the second extraction weight to serve as the extraction feature of the training data.
Specifically, the step of obtaining the original features of the training data may include:
acquiring a third extraction weight of the training data according to a preset third weight range; and acquiring data from the training data according to the third extraction weight to serve as the original features of the training data.
For ease of understanding, the process of extracting features of training data and obtaining raw features of training data will be described with a specific embodiment. Referring to fig. 2, W1 represents a weight coefficient of X entering the feature extraction layer (i.e., a first extraction weight of training data), W2 represents a weight coefficient of second feature data (a second extraction weight of the second feature data) obtained by extracting the first feature data (i.e., data entering the feature extraction layer in X) by the feature extraction layer, and W3 represents a weight coefficient of X performing a superposition operation with the extracted features (i.e., a third extraction weight of the training data).
The first extraction weight refers to a weight coefficient of training data entering the feature extraction layer. The second extraction weight is a weight coefficient of second feature data obtained by performing superposition operation on the original features. The third extraction weight is a weight coefficient of the training data subjected to the superposition operation with the extracted feature.
The first weight range refers to a value range of the first extraction weight. The second weight range refers to a value range of the second extraction weight. The third weight range refers to a value range of the third extraction weight.
The first feature data is data extracted from the training data based on the first extraction weight. The second feature data is data extracted from the first feature data by the feature extraction layer based on the second extraction weight.
The model to be trained respectively obtains a first extraction weight (namely, a specific value is obtained from the first weight range), a second extraction weight (namely, a specific value is obtained from the second weight range), and a third extraction weight (namely, a specific value is obtained from the second weight range) according to a preset first weight range, a preset second weight range, and combines the first extraction weight, the second extraction weight, and the third extraction weight to obtain various parameter value combinations, and the model to be trained conducts model training on each parameter value combination until the model converges to determine the final value of the model parameter. Specifically, after X (i.e., training data) is input to the feature characterization submodel, on one hand, data is obtained from X (i.e., training data) as first feature data by using a weight coefficient W1 (i.e., a first extraction weight), the first feature data is input to a feature extraction layer of the feature characterization submodel, and second feature data is extracted from the first feature data based on the feature extraction layer; then, data is acquired as an extracted feature from the second feature data by a weight coefficient W2 (i.e., a second extraction weight). On the other hand, with a weight coefficient W3 (i.e., a third extraction weight), data is acquired from X (i.e., training data) as an original feature.
In the embodiment of the application, W1, W2 and W3 are all model parameters, and the preset value ranges are all [0,1 ]; namely: the first extraction weight, the second extraction weight and the third extraction weight are model parameters, and the preset first weight range, the preset second weight range and the preset third weight range are all [0,1 ]. The model parameters to be trained can be automatically selected for combination by setting the minimum value of the model parameters to be 0 and the maximum value to be more than 0. When the value of the first extraction weight of the training data is 0 or the value of the second extraction weight of the first feature data is 0, the extracted feature of the training data is not considered by the model when the target feature is determined; when the value of the third extraction weight of the training data is 0, the model does not consider the original features of the training data when determining the target features. It is understood that, in other embodiments, the preset value ranges of W1, W2, and W3 may be set according to specific requirements.
In order to better understand how the model to be trained realizes the self-selection of the model parameters for combination, the following description is made in detail. Since the preset value ranges of W1, W2, and W3 are all [0,1], there may be three cases for the target feature learned by the information memory submodel:
1. w3 ═ 0. And after the input X is extracted by the feature extraction layer, the obtained extracted features are directly used as target features, and the information memory sub-model learns based on the target features. I.e. only the model parameters W1 and W2 are present.
2. W1 ═ 0 or W2 ═ 0. The input X is directly used as a target feature, and the information memorizing submodel learns based on the target feature. I.e. only the model parameters W3 are present.
3. W1! 0, W2! 0, W3! 0. Inputting X, firstly, extracting the X by a feature extraction layer to obtain an extraction feature; and then, extracting the features and performing feature superposition operation on the features and the X to obtain target features, and learning the information memory sub-model based on the target features. I.e. the presence model parameters W1, W2 and W3.
Therefore, by setting the minimum values of W1, W2 and W3 to 0, in the training process, the model to be trained can automatically select a parameter combination from three model parameters of W1, W2 and W3 to train the model. Namely, the model to be trained can automatically select a parameter combination from three model parameters, namely the first extraction weight, the second extraction weight and the third extraction weight, to carry out model training.
In the embodiment of the application, a value is obtained from the first weight range, the second weight range and the third weight range respectively and combined by presetting the first weight range, the second weight range and the third weight range, so that various different value combinations are formed; and the first weight range, the second weight range and the third weight range all contain the value 0, so that the number and the parameter value of the model parameters can be automatically determined, the number and the parameter value of the specific model parameters can be determined without manually selecting the model parameters, the model training process is completed, and the cost of manually adjusting the parameters is saved.
And B2, performing feature superposition on the original features and the extracted features through the feature characterization submodel to obtain the target features.
And C, acquiring the original features and the extracted features obtained in the step B2, and performing feature superposition operation (matrix addition operation and the like) on the original features and the extracted features based on the feature characterization submodel to finally obtain the target features.
In the embodiment of the application, the characteristics are extracted from the training data, the original characteristics of the training data are superposed with the extracted characteristics extracted from the training data to obtain the target characteristics with strong relevance to the abnormal transportation behaviors of the packages, and then the model to be trained learns the target characteristics to obtain a behavior recognition model; and then the behavior recognition model can extract characteristic data with strong relevance to the abnormal transportation behavior of the package from the acceleration data input to the behavior recognition model, and the behavior recognition model recognizes the abnormal transportation behavior of the package according to the extracted characteristic data, so that the accuracy of the behavior recognition model in recognizing the abnormal transportation behavior of the package is improved.
And 403, inputting the target characteristics into the information memory submodel of the model to be trained for learning, so as to obtain the behavior recognition model.
With reference to fig. 2, in order to enable the model to be trained to learn the target features extracted from the feature characterization submodel, and further form a behavior recognition model with a capability of recognizing abnormal transportation behaviors of packages, in this embodiment of the application, the model to be trained further includes an information memory submodel, and the information memory submodel is configured to learn the target features according to the abnormal transportation behavior labels. As an implementation, in the embodiment of the present application, the information memorizing submodel adopts an LSTM model.
Specifically, the target characteristics obtained in the step B2 are input into the information memorizing sub-model of the model to be trained, so that the information memorizing sub-model learns the target characteristics according to the abnormal transportation behavior label, and finally the behavior recognition model with the capability of recognizing the abnormal transportation behavior of the package is obtained.
In the embodiment of the application, a network structure of a model to be trained is provided with a feature characterization sub-model and an information memory sub-model, training data are characterized through the feature characterization sub-model, and then the characterized features are learned through the information memory sub-model, so that the model can extract the features for learning, and further the trained behavior recognition model is completed, and the accuracy of recognizing abnormal transportation behaviors is higher.
In order to better implement the abnormal transportation behavior method of the package in the embodiment of the application, on the basis of the abnormal transportation behavior method of the package, the embodiment of the application further provides a device for the abnormal transportation behavior of the package.
Referring to fig. 5, fig. 5 is a schematic structural diagram of an embodiment of the device for identifying abnormal transportation behavior of a package according to the present application.
In this application embodiment, the unusual behavior of transportation device of parcel includes:
an identification data acquisition unit 501, configured to acquire acceleration data of a package to be identified;
an input unit 502, configured to input the acceleration data into a trained behavior recognition model, where the acceleration data is used by the behavior recognition model to recognize motion accelerations of the parcel to be recognized in various directions, and the motion accelerations are used by the behavior recognition model to determine abnormal transportation behaviors of the parcel to be recognized;
the identifying unit 503 is configured to obtain an abnormal transportation behavior of the package to be identified, which is determined by the behavior identification model based on the acceleration data.
In some embodiments of the present application, the abnormal transportation behavior of the package further comprises:
a training data acquisition unit (not shown in the figure) for acquiring training data of the model to be trained;
and a model training unit (not shown in the figure) for inputting the training data into the model to be trained to learn so as to obtain the behavior recognition model.
In some embodiments of the present application, the training data obtaining unit is specifically configured to:
acquiring preset specific abnormal transportation behaviors and behavior segments of the specific abnormal transportation behaviors, and acquiring preset sampling duration and sampling rate of the behavior segments;
acquiring segment acceleration data of the behavior segment according to the sampling duration and the sampling rate;
and integrating the fragment acceleration data of all the behavior fragments of the specific abnormal transportation behavior into high-dimensional data serving as training data of a model to be trained.
In some embodiments of the present application, the model training unit is specifically configured to:
inputting the training data into a feature characterization submodel of the model to be trained;
acquiring target characteristics output by the characteristic characterization submodel based on the training data;
and inputting the target characteristics into an information memory sub-model of the model to be trained for learning to obtain the behavior recognition model.
In some embodiments of the present application, the model training unit is specifically configured to:
acquiring original features of the training data and extracted features of the training data through the feature characterization submodel;
and performing feature superposition on the original features and the extracted features through the feature characterization submodel to obtain the target features.
In some embodiments of the present application, the model training unit is specifically configured to:
acquiring a first extraction weight of the training data according to a preset first weight range;
acquiring first feature data from the training data according to the first extraction weight;
acquiring second characteristic data from the first characteristic data through a characteristic extraction layer of the characteristic characterization submodel;
acquiring a second extraction weight of the second feature data according to a preset second weight range;
and acquiring data from the second feature data according to the second extraction weight to serve as the extraction feature of the training data.
In some embodiments of the present application, the model training unit is specifically configured to:
acquiring a third extraction weight of the training data according to a preset third weight range;
and acquiring data from the training data according to the third extraction weight to serve as the original features of the training data.
The embodiment of the application further provides a terminal, which integrates the device for abnormal transportation behavior of any parcel provided by the embodiment of the application. As shown in fig. 6, it shows a schematic structural diagram of a terminal according to an embodiment of the present application, specifically:
the terminal may include components such as a processor 601 of one or more processing cores, memory 602 of one or more computer-readable storage media, a power supply 603, and an input unit 604. Those skilled in the art will appreciate that the terminal structure shown in fig. 6 is not intended to be limiting and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
Wherein:
the processor 601 is a control center of the terminal, connects various parts of the entire terminal using various interfaces and lines, and performs various functions of the terminal and processes data by operating or executing software programs and/or modules stored in the memory 602 and calling data stored in the memory 602, thereby performing overall monitoring of the terminal. Optionally, processor 601 may include one or more processing cores; in the embodiment of the present application, the processor 601 may integrate an application processor and a modem processor, wherein the application processor mainly processes an operating system, a user interface, an application program, and the like, and the modem processor mainly processes wireless communication. It will be appreciated that the modem processor described above may not be integrated into the processor 601.
The memory 602 may be used to store software programs and modules, and the processor 601 executes various functional applications and data processing by operating the software programs and modules stored in the memory 602. The memory 602 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data created according to the use of the terminal, and the like. Further, the memory 602 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device. Accordingly, the memory 602 may also include a memory controller to provide the processor 601 with access to the memory 602.
The terminal further includes a power supply 603 for supplying power to each component, and in this embodiment, the power supply 603 may be logically connected to the processor 601 through a power management system, so that functions of managing charging, discharging, power consumption management, and the like are implemented through the power management system. The power supply 603 may also include any component of one or more dc or ac power sources, recharging systems, power failure detection circuitry, power converters or inverters, power status indicators, and the like.
The terminal may further include an input unit 604, and the input unit 604 may be used to receive input numeric or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.
Although not shown, the terminal may further include a display unit and the like, which will not be described in detail herein. Specifically, in this embodiment, the processor 601 in the terminal loads the executable file corresponding to the process of one or more application programs into the memory 602 according to the following instructions, and the processor 601 runs the application programs stored in the memory 602, thereby implementing various functions as follows:
acquiring acceleration data of a parcel to be identified;
acquiring acceleration data of a parcel to be identified;
inputting the acceleration data into a trained behavior recognition model, wherein the acceleration data is used for the behavior recognition model to recognize the motion acceleration of the parcel to be recognized in each direction, and the motion acceleration is used for the behavior recognition model to determine the abnormal transportation behavior of the parcel to be recognized;
and acquiring the abnormal transportation behavior of the parcel to be identified, which is determined by the behavior identification model based on the acceleration data.
It will be understood by those skilled in the art that all or part of the steps of the methods of the above embodiments may be performed by instructions or by associated hardware controlled by the instructions, which may be stored in a computer readable storage medium and loaded and executed by a processor.
To this end, an embodiment of the present application provides a storage medium, which may include: read Only Memory (ROM), Random Access Memory (RAM), magnetic or optical disks, and the like. The computer program is loaded by a processor to execute the steps of any method for abnormal transportation behavior of packages provided by the embodiment of the application. For example, the computer program may be loaded by a processor to perform the steps of:
acquiring acceleration data of a parcel to be identified;
inputting the acceleration data into a trained behavior recognition model, wherein the acceleration data is used for the behavior recognition model to recognize the motion acceleration of the parcel to be recognized in each direction, and the motion acceleration is used for the behavior recognition model to determine the abnormal transportation behavior of the parcel to be recognized;
and acquiring the abnormal transportation behavior of the parcel to be identified, which is determined by the behavior identification model based on the acceleration data.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and parts that are not described in detail in a certain embodiment may refer to the above detailed descriptions of other embodiments, and are not described herein again.
In a specific implementation, each unit or structure may be implemented as an independent entity, or may be combined arbitrarily to be implemented as one or several entities, and the specific implementation of each unit or structure may refer to the foregoing method embodiment, which is not described herein again.
The above operations can be implemented in the foregoing embodiments, and are not described in detail herein.
The method, the device, the terminal and the storage medium for the abnormal transportation behavior of the package provided by the embodiment of the application are introduced in detail, a specific example is applied in the description to explain the principle and the implementation mode of the application, and the description of the embodiment is only used for helping to understand the method and the core idea of the application; meanwhile, for those skilled in the art, according to the idea of the embodiment of the present application, the specific implementation manner and the application range may be changed, and in summary, the content of the present specification should not be construed as a limitation to the present application.

Claims (10)

1. An abnormal transportation behavior identification method for a package is characterized by comprising the following steps:
acquiring acceleration data of a parcel to be identified;
inputting the acceleration data into a trained behavior recognition model, wherein the acceleration data is used for the behavior recognition model to recognize the motion acceleration of the parcel to be recognized in each direction, and the motion acceleration is used for the behavior recognition model to determine the abnormal transportation behavior of the parcel to be recognized;
and acquiring the abnormal transportation behavior of the parcel to be identified, which is determined by the behavior identification model based on the acceleration data.
2. The method for identifying abnormal transportation behavior of packages according to claim 1, wherein the inputting the acceleration data into the trained behavior identification model further comprises:
acquiring training data of a model to be trained;
and inputting the training data into the model to be trained for learning to obtain the behavior recognition model.
3. The method for identifying abnormal transportation behavior of package according to claim 2, wherein the obtaining training data of the model to be trained comprises:
acquiring preset specific abnormal transportation behaviors and behavior segments of the specific abnormal transportation behaviors, and acquiring preset sampling duration and sampling rate of the behavior segments;
acquiring segment acceleration data of the behavior segment according to the sampling duration and the sampling rate;
and integrating the fragment acceleration data of all the behavior fragments of the specific abnormal transportation behavior into high-dimensional data serving as training data of a model to be trained.
4. The method for identifying abnormal transportation behavior of package according to claim 2 or 3, wherein the inputting the training data into the model to be trained for learning to obtain the behavior identification model comprises:
inputting the training data into a feature characterization submodel of the model to be trained;
acquiring target characteristics output by the characteristic characterization submodel based on the training data;
and inputting the target characteristics into an information memory sub-model of the model to be trained for learning to obtain the behavior recognition model.
5. The method for identifying abnormal transportation behavior of packages according to claim 4, wherein the obtaining of the target feature of the feature characterization submodel based on the training data output comprises:
acquiring original features of the training data and extracted features of the training data through the feature characterization submodel;
and performing feature superposition on the original features and the extracted features through the feature characterization submodel to obtain the target features.
6. The method for identifying abnormal transportation behavior of packages according to claim 5, wherein the obtaining of the extracted features of the training data comprises:
acquiring a first extraction weight of the training data according to a preset first weight range;
acquiring first feature data from the training data according to the first extraction weight;
acquiring second characteristic data from the first characteristic data through a characteristic extraction layer of the characteristic characterization submodel;
acquiring a second extraction weight of the second feature data according to a preset second weight range;
and acquiring data from the second feature data according to the second extraction weight to serve as the extraction feature of the training data.
7. The method for identifying abnormal transportation behavior of packages according to claim 5, wherein the obtaining of the original features of the training data comprises:
acquiring a third extraction weight of the training data according to a preset third weight range;
and acquiring data from the training data according to the third extraction weight to serve as the original features of the training data.
8. An abnormal transportation behavior recognition apparatus for a package, characterized by comprising:
the identification data acquisition unit is used for acquiring acceleration data of the parcel to be identified;
the input unit is used for inputting the acceleration data into a trained behavior recognition model, the acceleration data is used for the behavior recognition model to recognize the motion acceleration of the parcel to be recognized in each direction, and the motion acceleration is used for the behavior recognition model to determine the abnormal transportation behavior of the parcel to be recognized;
and the identification unit is used for acquiring the abnormal transportation behavior of the parcel to be identified, which is determined by the behavior identification model based on the acceleration data.
9. An abnormal transportation behavior recognition terminal for a package, comprising: memory, processor and a package abnormal transportation behavior recognition program stored on the memory and executable on the processor, the package abnormal transportation behavior recognition program when executed by the processor implementing the steps of the package abnormal transportation behavior recognition method according to any one of claims 1 to 7.
10. A storage medium, characterized in that the storage medium has stored thereon an abnormal transportation behavior recognition program of a package, which when executed by a processor implements the steps of the abnormal transportation behavior recognition method of a package according to any one of claims 1 to 7.
CN201911108587.7A 2019-11-13 2019-11-13 Method and device for identifying abnormal transportation behavior of package, terminal and storage medium Pending CN112800814A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911108587.7A CN112800814A (en) 2019-11-13 2019-11-13 Method and device for identifying abnormal transportation behavior of package, terminal and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911108587.7A CN112800814A (en) 2019-11-13 2019-11-13 Method and device for identifying abnormal transportation behavior of package, terminal and storage medium

Publications (1)

Publication Number Publication Date
CN112800814A true CN112800814A (en) 2021-05-14

Family

ID=75803660

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911108587.7A Pending CN112800814A (en) 2019-11-13 2019-11-13 Method and device for identifying abnormal transportation behavior of package, terminal and storage medium

Country Status (1)

Country Link
CN (1) CN112800814A (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106897670A (en) * 2017-01-19 2017-06-27 南京邮电大学 A kind of express delivery violence sorting recognition methods based on computer vision
CN108830527A (en) * 2018-05-23 2018-11-16 深圳市恒兴泰胶粘制品有限公司 A kind of logistics recorder and its method for tracing
CN109726682A (en) * 2018-12-29 2019-05-07 南京信息工程大学 A kind of human motion recognition method towards weak label sensor data
US20190318202A1 (en) * 2016-10-31 2019-10-17 Tencent Technology (Shenzhen) Company Limited Machine learning model training method and apparatus, server, and storage medium

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190318202A1 (en) * 2016-10-31 2019-10-17 Tencent Technology (Shenzhen) Company Limited Machine learning model training method and apparatus, server, and storage medium
CN106897670A (en) * 2017-01-19 2017-06-27 南京邮电大学 A kind of express delivery violence sorting recognition methods based on computer vision
CN108830527A (en) * 2018-05-23 2018-11-16 深圳市恒兴泰胶粘制品有限公司 A kind of logistics recorder and its method for tracing
CN109726682A (en) * 2018-12-29 2019-05-07 南京信息工程大学 A kind of human motion recognition method towards weak label sensor data

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
陈飞 等人: ""基于CNN-LSTMs混合模型的人体行为识别方法"", 《信息技术与信息化》, vol. 2019, no. 04, pages 32 - 34 *

Similar Documents

Publication Publication Date Title
CN110555347B (en) Vehicle target identification method and device with dangerous cargo-carrying behavior and electronic equipment
JP2022521038A (en) Face recognition methods, neural network training methods, devices and electronic devices
CN102915372A (en) Image retrieval method, device and system
CN106250838A (en) vehicle identification method and system
CN107808126A (en) Vehicle retrieval method and device
CN101221623A (en) Object type on-line training and recognizing method and system thereof
CN110705489B (en) Training method and device for target recognition network, computer equipment and storage medium
CN104331691A (en) Vehicle logo classifier training method, vehicle logo recognition method and device
CN114937179B (en) Junk image classification method and device, electronic equipment and storage medium
CN109523793A (en) The methods, devices and systems of intelligent recognition information of vehicles
Hervieu et al. A statistical video content recognition method using invariant features on object trajectories
CN106384089A (en) Human body reliable detection method based on lifelong learning
CN112800814A (en) Method and device for identifying abnormal transportation behavior of package, terminal and storage medium
US11276285B2 (en) Artificial intelligence based motion detection
CN115298705A (en) License plate recognition method and device, electronic equipment and storage medium
CN114255435A (en) Method and device for detecting abnormality of transport device, electronic apparatus, and storage medium
CN112990245A (en) Article identification method, apparatus, device and storage medium
CN109542229B (en) Gesture recognition method, user equipment, storage medium and device
CN111382631B (en) Identification method, identification device, terminal, server and storage medium
CN113469994A (en) Pantograph detection method, pantograph detection device, electronic apparatus, and storage medium
CN114519793A (en) Target object detection method and device, electronic equipment and storage medium
CN112560685A (en) Facial expression recognition method and device and storage medium
KR20210041856A (en) Method and apparatus for generating learning data required to learn animation characters based on deep learning
CN102314612A (en) Method and device for identifying smiling face image and image acquisition equipment
CN111860661A (en) Data analysis method and device based on user behavior, electronic equipment and medium

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