CN114140751A - Examination room monitoring method and system - Google Patents
Examination room monitoring method and system Download PDFInfo
- Publication number
- CN114140751A CN114140751A CN202111516685.1A CN202111516685A CN114140751A CN 114140751 A CN114140751 A CN 114140751A CN 202111516685 A CN202111516685 A CN 202111516685A CN 114140751 A CN114140751 A CN 114140751A
- Authority
- CN
- China
- Prior art keywords
- training
- cheating
- image
- monitoring
- video
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 161
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000001514 detection method Methods 0.000 claims abstract description 121
- 230000003068 static effect Effects 0.000 claims abstract description 42
- 238000013528 artificial neural network Methods 0.000 claims abstract description 27
- 241000209202 Bromus secalinus Species 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 287
- 238000000605 extraction Methods 0.000 claims description 26
- 238000002372 labelling Methods 0.000 claims description 10
- 238000010586 diagram Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000006399 behavior Effects 0.000 abstract description 15
- 238000004590 computer program Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 206010000117 Abnormal behaviour Diseases 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an examination room monitoring method and system. And collecting a monitoring image. The monitoring images are monitoring images for monitoring all students in the examination room. And inputting the monitoring image into a cheating detection model, and judging whether a student cheats. And if the students cheat, acquiring cheating videos of 10 frames before and after the monitoring image. And sending the cheating video to a monitoring person. The cheating detection model comprises a cheating static detection layer, a cheating dynamic detection layer, a monitoring video acquisition layer and two full connection layers. The method for collecting the monitoring images is adopted to detect the cheating behaviors, so that the freedom and the accuracy of cheating judgment are greatly improved. The neural network is trained by judging the cheating type, and the accuracy of judging whether cheating occurs is improved. According to the related actions on the judgment time sequence, the accuracy of cheating judgment is greatly improved.
Description
Technical Field
The invention relates to the technical field of computers, in particular to an examination room monitoring method and system.
Background
At present, a plurality of methods are adopted for limiting cheating, such as the situation of adding a shielding device and the like. There is no other way to detect cheating behaviors, and at most, the behavior of adding invigilators is adopted to detect the cheating behaviors. The detection of the cheating behavior is difficult because the cheating can not be accurately detected due to personal reasons, such as the reason that the sight line can not be considered, through manual monitoring. And few methods for automatically detecting cheating are used for directly detecting the cheating, because the cheating behavior is not judged well, the cheating judgment accuracy is low, and the judgment is easy to be wrong.
Disclosure of Invention
The invention aims to provide an examination room monitoring method and system, which are used for solving the problems in the prior art
In a first aspect, an embodiment of the present invention provides an examination room monitoring method, including:
collecting a monitoring image; the monitoring images are monitoring images for monitoring all students in an examination room;
inputting the monitoring image into a cheating detection model, and judging whether a student cheats;
if the students cheat, acquiring cheating videos of 10 frames before and after the monitoring image; sending the cheating video to a monitoring person;
the cheating detection model comprises a cheating static detection layer, a cheating dynamic detection layer, a monitoring video acquisition layer and two full connection layers:
the input of the cheating static detection layer is a monitoring image; the input of the first full connection layer is the output of the cheating static detection layer; the output of the first full-connection layer is a first characteristic vector which is used for judging whether cheating is possible to happen or not; if the cheating is possible, determining the input of the monitoring video acquisition layer as a monitoring image; the input of the cheating dynamic detection layer is the output of the monitoring video acquisition layer; the input of the second full connection layer is the output of the cheating dynamic detection layer; the output of the second full-connection layer is a second feature vector which is used for judging whether the students cheat.
Optionally, the cheating detection model training method includes:
obtaining a training set, wherein the training set comprises a plurality of training images, a plurality of sections of training videos and labeling data; the training video represents a process of student cheating; the training images are images which can determine to indicate cheating in the training video; the annotation data comprises image annotation data and video annotation data; the image annotation data comprises annotation cheating conditions and annotation cheating types aiming at the images; the marked cheating condition indicates whether cheating action occurs or not; the video annotation data represents whether the students in the images cheat or not;
inputting the training image into a cheating static detection layer, and performing feature extraction to obtain a first training feature map;
inputting the first training feature map into a first full-connection layer to obtain a first training feature vector; the first training feature vector judges whether the behavior in the feature map is cheating or not; the first training feature vector comprises a cheating condition and a cheating type; the cheating condition represents whether cheating action occurs or not;
obtaining a first training loss value based on the first training feature vector and the labeling data;
inputting the first training video into a cheating dynamic detection layer, and performing feature extraction to obtain a second training feature map; the first training video is a training video containing the training image;
inputting the second training feature map into a second full-connection layer to obtain a second training feature vector; the second training feature vector judges whether the behavior in the feature map is cheating or not;
obtaining a second training loss value based on the second training feature vector and the labeled data;
obtaining a training loss value; the training loss value is the sum of the first training loss value and the second training loss value;
obtaining the current training iteration times of a detection model and the preset maximum iteration times of the training of the detection model;
and stopping training when the loss value is less than or equal to a threshold value or the training iteration number reaches the maximum iteration number, so as to obtain a trained detection model.
Optionally, the inputting the training image into the cheating static detection layer, and performing feature extraction to obtain a first training feature map includes:
obtaining a binaryzation training image; the binarization training image is an image obtained by binarization of the training image;
inputting the binaryzation training image into the cheating static detection layer, and performing feature extraction to obtain a first training feature map; and the cheating static detection layer is used for extracting the image characteristics of the binaryzation training image.
Optionally, the inputting the first training surveillance video into the cheating dynamic detection layer, and performing feature extraction to obtain a second training feature map, including:
inputting a first image of the first training monitoring video into a first neural network to obtain a first characteristic diagram; the first image is a first frame image of the first training surveillance video;
inputting the first feature map and a second image of the first training monitoring video into a second neural network to obtain a second feature map; the second image is separated from the first image by 5 frames;
inputting a current frame monitoring image of the first training monitoring video and a front monitoring image which is different from the current frame monitoring image by 5 frames into a corresponding neural network for multiple times to know that the last frame image of the first training monitoring video is reached;
obtaining a second training feature map; and the second feature map is obtained by inputting feature maps obtained by the last frame image and the previous frame image of the first training monitoring video into a corresponding neural network.
Optionally, obtaining a first training loss value based on the first training feature vector and the labeled data includes:
the first training loss value is obtained by the following formula calculation method:
wherein Loss is the first training Loss value; n is the number of all input pictures; y isiIndicating the marking cheating condition of the ith image; p is a radical ofiIndicating the predicted cheating condition of the ith image; k represents the number of cheat types;
indicating the kth cheating type of the ith image label;
indicating the kth type of cheating for the i-th image prediction.
Optionally, obtaining a second training loss value based on the second training feature vector and the labeled data includes:
the second training loss value is obtained by the following formula:
wherein Loss1 is the second training Loss value; m is the number of all input pictures; y isjIndicating the mark cheating condition of the jth image; p is a radical ofjIndicating the predicted cheating condition of the jth image.
In a second aspect, an embodiment of the present invention provides an examination room monitoring system, including:
an acquisition module: collecting a monitoring image; the monitoring images are monitoring images for monitoring all students in an examination room;
a cheating judgment module: inputting the monitoring image into a cheating detection model, and judging whether a student cheats;
the cheating video sending module: if the students cheat, acquiring cheating videos of 10 frames before and after the monitoring image; sending the cheating video to a monitoring person;
the cheating detection model comprises a cheating static detection layer, a cheating dynamic detection layer, a monitoring video acquisition layer and two full connection layers:
the input of the cheating static detection layer is a monitoring image; the input of the first full connection layer is the output of the cheating static detection layer; the output of the first full-connection layer is a first characteristic vector which is used for judging whether cheating is possible to happen or not; if the cheating is possible, determining the input of the monitoring video acquisition layer as a monitoring image; the input of the cheating dynamic detection layer is the output of the monitoring video acquisition layer; the input of the second full connection layer is the output of the cheating dynamic detection layer; the output of the second full-connection layer is a second feature vector which is used for judging whether the students cheat.
Optionally, the cheating detection model training method includes:
obtaining a training set, wherein the training set comprises a plurality of training images, a plurality of sections of training videos and labeling data; the training video represents a process of student cheating; the training images are images which can determine to indicate cheating in the training video; the annotation data comprises image annotation data and video annotation data; the image annotation data comprises annotation cheating conditions and annotation cheating types aiming at the images; the marked cheating condition indicates whether cheating action occurs or not; the video annotation data represents whether the students in the images cheat or not;
inputting the training image into a cheating static detection layer, and performing feature extraction to obtain a first training feature map;
inputting the first training feature map into a first full-connection layer to obtain a first training feature vector; the first training feature vector judges whether the behavior in the feature map is cheating or not; the first training feature vector comprises a cheating condition and a cheating type; the cheating condition represents whether cheating action occurs or not;
obtaining a first training loss value based on the first training feature vector and the labeling data;
inputting the first training video into a cheating dynamic detection layer, and performing feature extraction to obtain a second training feature map; the first training video is a training video containing the training image;
inputting the second training feature map into a second full-connection layer to obtain a second training feature vector; the second training feature vector judges whether the behavior in the feature map is cheating or not;
obtaining a second training loss value based on the second training feature vector and the labeled data;
obtaining a training loss value; the training loss value is the sum of the first training loss value and the second training loss value;
obtaining the current training iteration times of a detection model and the preset maximum iteration times of the training of the detection model;
and stopping training when the loss value is less than or equal to a threshold value or the training iteration number reaches the maximum iteration number, so as to obtain a trained detection model.
Optionally, the inputting the training image into the cheating static detection layer, and performing feature extraction to obtain a first training feature map includes:
obtaining a binaryzation training image; the binarization training image is an image obtained by binarization of the training image;
inputting the binaryzation training image into the cheating static detection layer, and performing feature extraction to obtain a first training feature map; and the cheating static detection layer is used for extracting the image characteristics of the binaryzation training image.
Optionally, the inputting the first training surveillance video into the cheating dynamic detection layer, and performing feature extraction to obtain a second training feature map, including:
inputting a first image of the first training monitoring video into a first neural network to obtain a first characteristic diagram; the first image is a first frame image of the first training surveillance video;
inputting the first feature map and a second image of the first training monitoring video into a second neural network to obtain a second feature map; the second image is separated from the first image by 5 frames;
inputting a current frame monitoring image of the first training monitoring video and a front monitoring image which is different from the current frame monitoring image by 5 frames into a corresponding neural network for multiple times to know that the last frame image of the first training monitoring video is reached;
obtaining a second training feature map; and the second feature map is obtained by inputting feature maps obtained by the last frame image and the previous frame image of the first training monitoring video into a corresponding neural network.
Compared with the prior art, the embodiment of the invention achieves the following beneficial effects:
collecting a monitoring image; the monitoring images are monitoring images for monitoring all students in an examination room; and inputting the monitoring image into a cheating detection model, and judging whether a student cheats. If the students cheat, acquiring cheating videos of 10 frames before and after the monitoring image; sending the cheating video to a monitoring person; the cheating detection model comprises a cheating static detection layer, a cheating dynamic detection layer, a monitoring video acquisition layer and two full connection layers: the input of the cheating static detection layer is a monitoring image; the input of the first full connection layer is the output of the cheating static detection layer; the output of the first full-connection layer is a first characteristic vector which is used for judging whether cheating is possible to happen or not; if the cheating is possible, determining the input of the monitoring video acquisition layer as a monitoring image; the input of the cheating dynamic detection layer is the output of the monitoring video acquisition layer; the input of the second full connection layer is the output of the cheating dynamic detection layer; the output of the second full-connection layer is a second feature vector which is used for judging whether the students cheat.
The method for collecting the monitoring images is adopted to detect the cheating behaviors, so that the freedom and the accuracy of cheating judgment are greatly improved. Firstly, cheating judgment is carried out on the monitoring image, and whether cheating is carried out is judged. Meanwhile, the cheating type is judged, the neural network is trained through the judgment of the cheating type, and the accuracy of judging whether cheating occurs is improved. Meanwhile, a cheating dynamic detection layer is added for judging, namely whether cheating is performed or not is judged by utilizing actions expressed by multi-frame images in a section of video instead of cheating action images. According to the related actions on the judgment time sequence, the accuracy of cheating judgment is greatly improved.
Drawings
Fig. 1 is a flowchart of an examination room monitoring method according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of a test room monitoring method detection and training process according to an embodiment of the present invention.
Fig. 3 is a schematic structural diagram of an RNN cycle network according to an embodiment of the present invention.
Fig. 4 is a schematic block structure diagram of an electronic device according to an embodiment of the present invention.
The labels in the figure are: a bus 500; a receiver 501; a processor 502; a transmitter 503; a memory 504; a bus interface 505.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings.
Example 1
As shown in fig. 1, an embodiment of the present invention provides an examination room monitoring method, where the method includes:
s101: and collecting a monitoring image. The monitoring images are monitoring images for monitoring all students in the examination room.
S102: inputting the monitoring image into a cheating detection model, and judging whether a student cheats;
s103: and if the students cheat, acquiring cheating videos of 10 frames before and after the monitoring image. And sending the cheating video to a monitoring person.
The cheating detection model comprises a cheating static detection layer, a cheating dynamic detection layer, a monitoring video acquisition layer and two full connection layers:
the input of the cheating static detection layer is a monitoring image; the input of the first full connection layer is the output of the cheating static detection layer; the output of the first full-connection layer is a first characteristic vector which is used for judging whether cheating is possible to happen or not; if the cheating is possible, determining the input of the monitoring video acquisition layer as a monitoring image; the input of the cheating dynamic detection layer is the output of the monitoring video acquisition layer; the input of the second full connection layer is the output of the cheating dynamic detection layer; the output of the second full-connection layer is a second feature vector which is used for judging whether the students cheat.
In this embodiment, the output of the cheating monitoring video layer is a plurality of front and back monitoring images. The multiple front and rear monitoring images are 2 images acquired every 5 frames ahead of the time sequence of the monitoring images and 2 images acquired every 5 frames behind the time sequence of the monitoring images, and the monitoring images are driven to obtain the monitoring video.
Wherein, the position in the monitoring image can be obtained by determining the position of each student, thereby carrying out cheating identification on single students. If cheating is possible, a single student monitoring video is obtained. The single student surveillance video representation only comprises a plurality of continuous surveillance images of students before and after the cheating surveillance image. And inputting the single student monitoring video into a cheating dynamic detection layer, and judging whether cheating occurs or not.
Wherein, the detection and training process of the cheating detection model is as shown in fig. 2.
By the method, the basic action of the student in the action movement process can be reserved by acquiring the image every 5 frames, the calculation degree can be reduced, and the risk of gradient explosion is reduced.
Optionally, the cheating detection model training method includes:
obtaining a training set, wherein the training set comprises a plurality of training images, a plurality of sections of training videos and labeling data; the training video represents a process of student cheating; the training images are images which can determine to indicate cheating in the training video; the annotation data comprises image annotation data and video annotation data; the image annotation data comprises annotation cheating conditions and annotation cheating types aiming at the images; the marked cheating condition indicates whether cheating action occurs or not; the video annotation data indicates whether the students in the images cheat.
And inputting the training image into a cheating static detection layer, and performing feature extraction to obtain a first training feature map.
And inputting the first training feature map into a first full-connection layer to obtain a first training feature vector. And the first training feature vector judges whether the behavior in the feature map is cheating or not. The first training feature vector comprises a cheating condition and a cheating type; the cheating condition indicates whether a cheating action occurs.
Obtaining a first training loss value based on the first training feature vector and the labeling data;
inputting the first training video into a cheating dynamic detection layer, and performing feature extraction to obtain a second training feature map; the first training video is a training video containing the training image;
and inputting the second training feature map into a second full-connection layer to obtain a second training feature vector. And the second training feature vector judges whether the behavior in the feature map is cheating or not.
Obtaining a second training loss value based on the second training feature vector and the labeled data;
obtaining a training loss value; the training loss value is the sum of the first training loss value and the second training loss value;
and obtaining the current training iteration times of the detection model and the preset maximum iteration times of the training of the detection model.
And stopping training when the loss value is less than or equal to a threshold value or the training iteration number reaches the maximum iteration number, so as to obtain a trained detection model.
Wherein the threshold is 1.
By the method, the first training loss value and the second training loss value are obtained, and the neural network is trained according to a method of summing the first training loss value and the second training loss value to obtain a total loss value. So that the neural network as a whole can recognize whether cheating is occurring. And whether the cheating and the cheating type are judged on the cheating static detection layer, and whether the cheating is needed to be judged on the cheating dynamic detection layer only, so that on one hand, the cheating type is judged simply and conveniently by the cheating static detection layer. And secondly, the judgment of the cheating type is added to increase the accuracy of judging whether the cheating is carried out by the neural network, and actually, the function of the neural network is to judge whether the cheating is carried out.
Optionally, the method includes inputting the training image into a cheating static detection layer, and performing feature extraction to obtain a first training feature map, including:
and obtaining a binary training image. The binarization training image is an image obtained by binarization of the training image.
And inputting the binary training image into the cheating static detection layer, and performing feature extraction to obtain a first training feature map. And the cheating static detection layer is used for extracting the image characteristics of the binaryzation training image.
The cheating static detection layer comprises a plurality of convolution layers, a Resnet50 residual error network is adopted in the invention, and the structure and the output of the last three layers of the detection main network are shown in the following table 1:
TABLE 1
By the above method, the features of the binarized image are obtained. Firstly, the image is binarized, only the shape characteristics of the image are left, and other influence factors such as clothing, color and the like are removed, so that the shape-based characteristics of the image can be extracted more quickly and accurately. Meanwhile, the Resnet50 residual network used in the embodiment can be trained by a deeper network.
Optionally, the inputting the first training surveillance video into the cheating dynamic detection layer, and performing feature extraction to obtain a second training feature map, including:
inputting a first image of the first training monitoring video into a first neural network to obtain a first characteristic diagram; the first image is a first frame image of the first training surveillance video.
Inputting the first feature map and a second image of the first training monitoring video into a second neural network to obtain a second feature map; the second image is spaced 5 frames from the first image.
The last frame image of the first training surveillance video is known to be reached by inputting the current frame surveillance image of the first training surveillance video and the previous surveillance image which is 5 frames different from the current frame surveillance image into the corresponding neural network for a plurality of times.
Obtaining a second training feature map; and the second feature map is obtained by inputting feature maps obtained by the last frame image and the previous frame image of the first training monitoring video into a corresponding neural network.
Fig. 3 shows a structure of an image input RNN loop network representing 5 frames apart in a surveillance video in the present embodiment.
By the method, the cheating dynamic detection layer is of an RNN structure, and behavior judgment can be performed on the whole video by combining the feature map of the last video in the video. RNN relation capable of relating two picture characteristics
Optionally, obtaining a first training loss value based on the first training feature vector and the labeled data includes:
the first training loss value is obtained by the following formula calculation method:
wherein Loss is the first training Loss value. N is the number of all input pictures. y isiIndicating the marking cheating condition of the ith image. p is a radical ofiShowing the predicted cheating condition of the ith image. K represents the number of cheat types.
Indicating the k-th type of cheating on the ith image annotation.
Indicating the kth type of cheating for the i-th image prediction.
By the method, the loss of the cheating condition is firstly obtained, if the cheating condition is marked to be no cheating, the cheating type loss is not too useful, and therefore the cheating condition is marked in front of the cheating type loss by adding entropy. This increases the accuracy of the neural network in determining whether to cheat.
Optionally, obtaining a second training loss value based on the second training feature vector and the labeled data includes:
the second training loss value is obtained by the following formula:
wherein Loss1 is the second training Loss value. M is the number of all input pictures. y isjIndicating the label cheating condition of the jth image. p is a radical ofjIndicating the predicted cheating condition of the jth image.
By the above method, a second training loss value is obtained. And the second training loss value is obtained after classification prediction is carried out on the features extracted by the cheating dynamic detection layer. The condition of adding the cheating dynamic detection layer greatly increases the accuracy of action recognition in a period of time.
By the method, the monitoring image is input into the cheating detection model to detect the abnormal behavior. The abnormal behavior is a marked cheating behavior. The feature extraction is carried out in a convolution mode and then the detection is carried out, whether the cheating is an important result is detected, and which cheating type is only used for increasing the accuracy of detecting whether the cheating is carried out. Therefore, when the loss function is set, the loss function of judging the cheating type is multiplied by marking whether the data is cheated or not. The judgment of the cheating type is carried out if the cheating is carried out, and the judgment is not carried out continuously if the cheating is not carried out. Meanwhile, a cheating dynamic detection layer is added to increase the accuracy of cheating judgment, and because the RNN adopted by the cheating dynamic detection layer is a circulating network, whether cheating occurs or not is judged by judging the action relationship within a period of time, and the judgment is carried out according to the action relationship in time, so that the accuracy of judging whether cheating occurs or not is greatly increased.
Based on the examination room monitoring method, the embodiment of the invention also provides an examination room monitoring system, which comprises an acquisition module, a cheating judgment module and a cheating video sending module.
The acquisition module is used for acquiring the monitoring graph. The monitoring images are monitoring images for monitoring all students in the examination room.
The cheating judging module is used for inputting the monitoring image into the cheating detection model and judging whether a student cheats or not.
And the cheating video sending module obtains cheating videos of 10 frames before and after the monitoring image when students cheat. And send the cheating video to the monitoring personnel.
The specific manner in which the respective modules perform operations has been described in detail in the embodiments related to the method, and will not be elaborated upon here.
An embodiment of the present invention further provides an electronic device, as shown in fig. 4, including a memory 504, a processor 502, and a computer program stored on the memory 504 and executable on the processor 502, where the processor 502 implements the steps of any one of the above-described examination room monitoring methods when executing the program.
Where in fig. 4 a bus architecture (represented by bus 500) is shown, bus 500 may include any number of interconnected buses and bridges, and bus 500 links together various circuits including one or more processors, represented by processor 502, and memory, represented by memory 504. The bus 500 may also link together various other circuits such as peripherals, voltage regulators, power management circuits, and the like, which are well known in the art, and therefore, will not be described any further herein. A bus interface 505 provides an interface between the bus 500 and the receiver 501 and transmitter 503. The receiver 501 and the transmitter 503 may be the same element, i.e. a transceiver, providing a means for communicating with various other apparatus over a transmission medium. The processor 502 is responsible for managing the bus 500 and general processing, and the memory 504 may be used for storing data used by the processor 502 in performing operations.
Embodiments of the present invention further provide a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the steps of any one of the above-mentioned examination room monitoring methods and the above-mentioned related data.
The algorithms and displays presented herein are not inherently related to any particular computer, virtual machine, or other apparatus. Various general purpose systems may also be used with the teachings herein. The required structure for constructing such a system will be apparent from the description above. Moreover, the present invention is not directed to any particular programming language. It is appreciated that a variety of programming languages may be used to implement the teachings of the present invention as described herein, and any descriptions of specific languages are provided above to disclose the best mode of the invention.
In the description provided herein, numerous specific details are set forth. It is understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.
Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be interpreted as reflecting an intention that: that the invention as claimed requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules in the device in an embodiment may be adaptively changed and disposed in one or more devices different from the embodiment. The modules or units or components of the embodiments may be combined into one module or unit or component, and furthermore they may be divided into a plurality of sub-modules or sub-units or sub-components. All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or elements of any method or apparatus so disclosed, may be combined in any combination, except combinations where at least some of such features and/or processes or elements are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Furthermore, those skilled in the art will appreciate that while some embodiments herein include some features included in other embodiments, rather than other features, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments may be used in any combination.
The various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that a microprocessor or Digital Signal Processor (DSP) may be used in practice to implement some or all of the functionality of some or all of the components in an apparatus according to an embodiment of the invention. The present invention may also be embodied as apparatus or device programs (e.g., computer programs and computer program products) for performing a portion or all of the methods described herein. Such programs implementing the present invention may be stored on computer-readable media or may be in the form of one or more signals. Such a signal may be downloaded from an internet website or provided on a carrier signal or in any other form.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names.
Claims (10)
1. An examination room monitoring method is characterized by comprising the following steps:
collecting a monitoring image; the monitoring images are monitoring images for monitoring all students in an examination room;
inputting the monitoring image into a cheating detection model, and judging whether a student cheats;
if the students cheat, acquiring cheating videos of 10 frames before and after the monitoring image; sending the cheating video to a monitoring person;
the cheating detection model comprises a cheating static detection layer, a cheating dynamic detection layer, a monitoring video acquisition layer and two full connection layers:
the input of the cheating static detection layer is a monitoring image; the input of the first full connection layer is the output of the cheating static detection layer; the output of the first full-connection layer is a first characteristic vector which is used for judging whether cheating is possible to happen or not; if the cheating is possible, determining the input of the monitoring video acquisition layer as a monitoring image; the input of the cheating dynamic detection layer is the output of the monitoring video acquisition layer; the input of the second full connection layer is the output of the cheating dynamic detection layer; the output of the second full-connection layer is a second feature vector which is used for judging whether the students cheat.
2. The examination room monitoring method of claim 1, wherein the cheating detection model training method comprises the following steps:
obtaining a training set, wherein the training set comprises a plurality of training images, a plurality of sections of training videos and labeling data; the training video represents a process of student cheating; the training images are images which can determine to indicate cheating in the training video; the annotation data comprises image annotation data and video annotation data; the image annotation data comprises annotation cheating conditions and annotation cheating types aiming at the images; the marked cheating condition indicates whether cheating action occurs or not; the video annotation data represents whether the students in the images cheat or not;
inputting the training image into a cheating static detection layer, and performing feature extraction to obtain a first training feature map;
inputting the first training feature map into a first full-connection layer to obtain a first training feature vector; the first training feature vector judges whether the behavior in the feature map is cheating or not; the first training feature vector comprises a cheating condition and a cheating type; the cheating condition represents whether cheating action occurs or not;
obtaining a first training loss value based on the first training feature vector and the labeling data;
inputting the first training video into a cheating dynamic detection layer, and performing feature extraction to obtain a second training feature map; the first training video is a training video containing the training image;
inputting the second training feature map into a second full-connection layer to obtain a second training feature vector; the second training feature vector judges whether the behavior in the feature map is cheating or not;
obtaining a second training loss value based on the second training feature vector and the labeled data;
obtaining a training loss value; the training loss value is the sum of the first training loss value and the second training loss value;
obtaining the current training iteration times of a detection model and the preset maximum iteration times of the training of the detection model;
and stopping training when the loss value is less than or equal to a threshold value or the training iteration number reaches the maximum iteration number, so as to obtain a trained detection model.
3. The examination room monitoring method according to claim 2, wherein the step of inputting the training image into a cheating static detection layer for feature extraction to obtain a first training feature map comprises:
obtaining a binaryzation training image; the binarization training image is an image obtained by binarization of the training image;
inputting the binaryzation training image into the cheating static detection layer, and performing feature extraction to obtain a first training feature map; and the cheating static detection layer is used for extracting the image characteristics of the binaryzation training image.
4. The examination room monitoring method according to claim 2, wherein the step of inputting the first training surveillance video into a cheating dynamic detection layer for feature extraction to obtain a second training feature map comprises:
inputting a first image of the first training monitoring video into a first neural network to obtain a first characteristic diagram; the first image is a first frame image of the first training surveillance video;
inputting the first feature map and a second image of the first training monitoring video into a second neural network to obtain a second feature map; the second image is separated from the first image by 5 frames;
inputting a current frame monitoring image of the first training monitoring video and a front monitoring image which is different from the current frame monitoring image by 5 frames into a corresponding neural network for multiple times to know that the last frame image of the first training monitoring video is reached;
obtaining a second training feature map; and the second feature map is obtained by inputting feature maps obtained by the last frame image and the previous frame image of the first training monitoring video into a corresponding neural network.
5. The examination room monitoring method according to claim 2, wherein the obtaining a first training loss value based on the first training eigenvector and the labeled data comprises:
the first training loss value is obtained by the following formula calculation method:
wherein Loss is the first training Loss value; n is the number of all input pictures; y isiIndicating the marking cheating condition of the ith image; p is a radical ofiIndicating the predicted cheating condition of the ith image; k represents the number of cheat types;
indicating the kth cheating type of the ith image label;
indicating the kth type of cheating for the i-th image prediction.
6. The examination room monitoring method according to claim 2, wherein the obtaining a second training loss value based on the second training eigenvector and the labeled data comprises:
the second training loss value is obtained by the following formula:
wherein Loss1 is the second training Loss value; m is the number of all input pictures; y isjIndicating the mark cheating condition of the jth image; p is a radical ofjIndicating the predicted cheating condition of the jth image.
7. An examination room monitoring system, comprising:
an acquisition module: collecting a monitoring image; the monitoring images are monitoring images for monitoring all students in an examination room;
a cheating judgment module: inputting the monitoring image into a cheating detection model, and judging whether a student cheats;
the cheating video sending module: if the students cheat, acquiring cheating videos of 10 frames before and after the monitoring image; sending the cheating video to a monitoring person;
the cheating detection model comprises a cheating static detection layer, a cheating dynamic detection layer, a monitoring video acquisition layer and two full connection layers:
the input of the cheating static detection layer is a monitoring image; the input of the first full connection layer is the output of the cheating static detection layer; the output of the first full-connection layer is a first characteristic vector which is used for judging whether cheating is possible to happen or not; if the cheating is possible, determining the input of the monitoring video acquisition layer as a monitoring image; the input of the cheating dynamic detection layer is the output of the monitoring video acquisition layer; the input of the second full connection layer is the output of the cheating dynamic detection layer; the output of the second full-connection layer is a second feature vector which is used for judging whether the students cheat.
8. An examination room monitoring system according to claim 7, wherein the cheating detection model training method comprises:
obtaining a training set, wherein the training set comprises a plurality of training images, a plurality of sections of training videos and labeling data; the training video represents a process of student cheating; the training images are images which can determine to indicate cheating in the training video; the annotation data comprises image annotation data and video annotation data; the image annotation data comprises annotation cheating conditions and annotation cheating types aiming at the images; the marked cheating condition indicates whether cheating action occurs or not; the video annotation data represents whether the students in the images cheat or not;
inputting the training image into a cheating static detection layer, and performing feature extraction to obtain a first training feature map;
inputting the first training feature map into a first full-connection layer to obtain a first training feature vector; the first training feature vector judges whether the behavior in the feature map is cheating or not; the first training feature vector comprises a cheating condition and a cheating type; the cheating condition represents whether cheating action occurs or not;
obtaining a first training loss value based on the first training feature vector and the labeling data;
inputting the first training video into a cheating dynamic detection layer, and performing feature extraction to obtain a second training feature map; the first training video is a training video containing the training image;
inputting the second training feature map into a second full-connection layer to obtain a second training feature vector; the second training feature vector judges whether the behavior in the feature map is cheating or not;
obtaining a second training loss value based on the second training feature vector and the labeled data;
obtaining a training loss value; the training loss value is the sum of the first training loss value and the second training loss value;
obtaining the current training iteration times of a detection model and the preset maximum iteration times of the training of the detection model;
and stopping training when the loss value is less than or equal to a threshold value or the training iteration number reaches the maximum iteration number, so as to obtain a trained detection model.
9. The examination room monitoring system of claim 8, wherein the inputting the training image into a cheating static detection layer for feature extraction to obtain a first training feature map comprises:
obtaining a binaryzation training image; the binarization training image is an image obtained by binarization of the training image;
inputting the binaryzation training image into the cheating static detection layer, and performing feature extraction to obtain a first training feature map; and the cheating static detection layer is used for extracting the image characteristics of the binaryzation training image.
10. The examination room monitoring system of claim 8, wherein the inputting the first training surveillance video into the cheating dynamic detection layer for feature extraction to obtain a second training feature map comprises:
inputting a first image of the first training monitoring video into a first neural network to obtain a first characteristic diagram; the first image is a first frame image of the first training surveillance video;
inputting the first feature map and a second image of the first training monitoring video into a second neural network to obtain a second feature map; the second image is separated from the first image by 5 frames;
inputting a current frame monitoring image of the first training monitoring video and a front monitoring image which is different from the current frame monitoring image by 5 frames into a corresponding neural network for multiple times to know that the last frame image of the first training monitoring video is reached;
obtaining a second training feature map; and the second feature map is obtained by inputting feature maps obtained by the last frame image and the previous frame image of the first training monitoring video into a corresponding neural network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111516685.1A CN114140751B (en) | 2021-12-13 | 2021-12-13 | Examination room monitoring method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111516685.1A CN114140751B (en) | 2021-12-13 | 2021-12-13 | Examination room monitoring method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114140751A true CN114140751A (en) | 2022-03-04 |
| CN114140751B CN114140751B (en) | 2024-02-09 |
Family
ID=80382012
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111516685.1A Active CN114140751B (en) | 2021-12-13 | 2021-12-13 | Examination room monitoring method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114140751B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116153312A (en) * | 2023-03-05 | 2023-05-23 | 广州网才信息技术有限公司 | Online pen test method and device using voice recognition |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107958235A (en) * | 2017-12-28 | 2018-04-24 | 泰康保险集团股份有限公司 | A kind of facial image detection method, device, medium and electronic equipment |
| CN109829392A (en) * | 2019-01-11 | 2019-05-31 | 平安科技（深圳）有限公司 | Examination hall cheating recognition methods, system, computer equipment and storage medium |
| CN111259844A (en) * | 2020-01-21 | 2020-06-09 | 陕西师范大学 | Real-time monitoring method for examinees in standardized examination room |
| CN111598049A (en) * | 2020-05-29 | 2020-08-28 | 中国工商银行股份有限公司 | Cheating recognition method and apparatus, electronic device, and medium |
| CN112036299A (en) * | 2020-08-31 | 2020-12-04 | 山东科技大学 | Examination cheating behavior detection method and system under standard examination room environment |
| CN112417989A (en) * | 2020-10-30 | 2021-02-26 | 四川天翼网络服务有限公司 | Invigilator violation identification method and system |
| CN112492343A (en) * | 2020-12-16 | 2021-03-12 | 浙江大华技术股份有限公司 | Video live broadcast monitoring method and related device |
| CA3180213A1 (en) * | 2020-04-17 | 2021-10-21 | Bayer Cropscience Lp | Image monitoring for control of invasive grasses |
-
2021
- 2021-12-13 CN CN202111516685.1A patent/CN114140751B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107958235A (en) * | 2017-12-28 | 2018-04-24 | 泰康保险集团股份有限公司 | A kind of facial image detection method, device, medium and electronic equipment |
| CN109829392A (en) * | 2019-01-11 | 2019-05-31 | 平安科技（深圳）有限公司 | Examination hall cheating recognition methods, system, computer equipment and storage medium |
| CN111259844A (en) * | 2020-01-21 | 2020-06-09 | 陕西师范大学 | Real-time monitoring method for examinees in standardized examination room |
| CA3180213A1 (en) * | 2020-04-17 | 2021-10-21 | Bayer Cropscience Lp | Image monitoring for control of invasive grasses |
| CN111598049A (en) * | 2020-05-29 | 2020-08-28 | 中国工商银行股份有限公司 | Cheating recognition method and apparatus, electronic device, and medium |
| CN112036299A (en) * | 2020-08-31 | 2020-12-04 | 山东科技大学 | Examination cheating behavior detection method and system under standard examination room environment |
| CN112417989A (en) * | 2020-10-30 | 2021-02-26 | 四川天翼网络服务有限公司 | Invigilator violation identification method and system |
| CN112492343A (en) * | 2020-12-16 | 2021-03-12 | 浙江大华技术股份有限公司 | Video live broadcast monitoring method and related device |
Non-Patent Citations (2)
| Title |
|---|
| HAMZA RIAZ 等: "ANOMALOUS HUMAN ACTION DETECTION USING A CASCADE OF DEEP LEARNING MODELS", 《2021 9TH EUROPEAN WORKSHOP ON VISUAL INFORMATION PROCESSING》, pages 1 - 5 * |
| 万子云 等: "基于深度学习的MOOC 作弊行为检测研究", 《信息安全学报》, vol. 6, no. 1, pages 32 - 39 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116153312A (en) * | 2023-03-05 | 2023-05-23 | 广州网才信息技术有限公司 | Online pen test method and device using voice recognition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114140751B (en) | 2024-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110705405B (en) | Target labeling method and device | |
| TWI716012B (en) | Sample labeling method, device, storage medium and computing equipment, damage category identification method and device | |
| CN110264444B (en) | Damage detection method and device based on weak segmentation | |
| CN110569703B (en) | Computer-implemented method and device for identifying damage from picture | |
| WO2020134102A1 (en) | Article recognition method and device, vending system, and storage medium | |
| US10671887B2 (en) | Best image crop selection | |
| CN108337505A (en) | Information acquisition method and device | |
| CN110660078B (en) | Object tracking method, device, computer equipment and storage medium | |
| AU2021203869B2 (en) | Methods, devices, electronic apparatuses and storage media of image processing | |
| WO2024060684A1 (en) | Model training method, image processing method, device, and storage medium | |
| CN111814690A (en) | Target re-identification method and device and computer readable storage medium | |
| CN111753746A (en) | Attribute recognition model training method, recognition method, electronic device, and storage medium | |
| CN103049747B (en) | The human body image utilizing the colour of skin knows method for distinguishing again | |
| CN116934195A (en) | Commodity information checking method and device, electronic equipment and storage medium | |
| CN115082752A (en) | Target detection model training method, device, equipment and medium based on weak supervision | |
| CN114140751B (en) | Examination room monitoring method and system | |
| CN106327531A (en) | Panorama video identification method and device, and video playing method and device | |
| CN109711434B (en) | Method, apparatus, device and medium for acquiring and evaluating VQA system training data | |
| CN108549899B (en) | Image identification method and device | |
| CN113435353A (en) | Multi-mode-based in-vivo detection method and device, electronic equipment and storage medium | |
| CN113255766B (en) | Image classification method, device, equipment and storage medium | |
| CN115438945A (en) | Risk identification method, device, equipment and medium based on power equipment inspection | |
| CN111860357B (en) | Attendance rate calculating method and device based on living body identification, terminal and storage medium | |
| CN114359700A (en) | Data processing method and device, electronic equipment and storage medium | |
| CN106446902B (en) | Non-legible image-recognizing method and device |
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 |