CN117523631B - Image-based bid evaluation room early warning method - Google Patents

Abstract

The invention relates to an image-based bid evaluation room early warning method, which comprises the following steps: acquiring an image of a bid evaluation room in real time between the beginning and the end of bid evaluation; based on the image, each time it is determined that personnel are present at the entrance of the rating room: recording the appearance time and tracking the appearance personnel; determining whether a human face appears according to the tracking result; if the face is still not recognized when the appearance time +alpha is reached, the identity confirmation is not passed; if the face is recognized within the occurrence time +alpha, the face recognition is carried out on the recognized face, if the face recognition result is that the person appearing is a label evaluation expert, the identity confirmation is passed, and if the face recognition result is that the person appearing is not a label evaluation expert, the identity confirmation is not passed; if the identity is not confirmed, early warning is carried out; if the identity is confirmed to pass, the personnel appearing are tracked in real time according to the image, and early warning is carried out according to the identification of the personnel appearing and the tracking result, so that the automatic early warning of the label evaluation room is realized.

Description

Image-based bid evaluation room early warning method

Technical Field

The invention relates to the technical field of image processing, in particular to an image-based marking room early warning method.

Background

The bid evaluation refers to the act of the bid evaluation committee and the signer to review, evaluate and compare the bid documents according to the bid evaluation standards and methods specified by the bid documents.

The bid evaluation is an important stage in bidding and bidding activities, and whether the bid evaluation really achieves disclosure, fairness and fairness determines whether the whole bidding and bidding activities are fairness and fairness; the quality of the bid evaluation determines whether a winning bid candidate which best meets the requirements of the bid-drawing project can be selected from a plurality of bidding competitors.

Therefore, the early warning of abnormal conditions (such as intrusion of unauthorized personnel, equipment abnormality and the like) is very important in the evaluation process.

The existing scheme carries out early warning through a mode of manually checking the monitoring video, and the mode not only consumes labor, but also can have the condition of abnormal missing report caused by personnel errors.

Disclosure of Invention

First, the technical problem to be solved

In view of the above-mentioned drawbacks and shortcomings of the prior art, the present invention provides an image-based warning method for a marking room.

(II) technical scheme

In order to achieve the above purpose, the main technical scheme adopted by the invention comprises the following steps:

s101, acquiring an image of a bid evaluation room in real time between the beginning and the end of bid evaluation;

S102, based on the image, each time the personnel at the entrance of the evaluation room are determined to appear, the following steps are executed:

determining a pre-allocation identifier, and setting the identifier of the personnel appearing as the pre-allocation identifier;

recording the appearance time and tracking the appearance personnel;

determining whether a human face appears according to the tracking result;

If the face is still not recognized when the appearance time +alpha is reached, the identity confirmation is not passed; if the face is recognized within the occurrence time +alpha, the face recognition is carried out on the recognized face, if the face recognition result is that the person appearing is a label evaluation expert, the identity confirmation is passed, and if the face recognition result is that the person appearing is not a label evaluation expert, the identity confirmation is not passed; wherein alpha is a preset duration threshold;

If the identity is not confirmed, early warning is carried out; if the identity confirmation is passed, the identification of the personnel appearing is updated to be the identification of the evaluation expert, the personnel appearing is tracked in real time according to the image, and early warning is carried out according to the identification of the personnel appearing and the tracking result.

Optionally, early warning is performed according to the identification and tracking result of the personnel, including:

determining whether the personnel enter a marking room for the first time according to the identification of the personnel;

If the person enters the label evaluation room for the first time, before the label evaluation is finished, determining that the person leaves the label evaluation room according to the tracking result, storing the leaving time by taking the mark of the person as the mark, calculating a leaving time threshold beta, and storing beta; if the departure time is +beta, the personnel do not enter the evaluation room again, and early warning is carried out.

Optionally, calculating the departure time period threshold β includes:

Acquiring whether the personnel present submit the bid evaluation result;

If the personnel present have submitted the bid evaluation result, calculating a first difference value = bid evaluation ending time-leaving time; when the first difference value is not greater than a preset minimum threshold value, determining beta as a preset maximum value; when the first difference is greater than a preset minimum threshold, β is determined as the minimum threshold [1+ (total time length of rating-first difference)/total time length of rating ].

Optionally, after obtaining whether the personnel present submit the bid evaluation result, the method further comprises:

If the personnel do not submit the marking result, determining that beta is 0 when the first difference value is not greater than a preset minimum threshold value; and when the first difference value is larger than a preset minimum threshold value, determining beta as the minimum threshold value.

Optionally, after determining whether the person is first entering the rating room, the method further includes:

if the person enters the label evaluation room for the first time, acquiring the last leaving time according to the mark of the person;

calculate a second difference = time of occurrence-last time of departure;

If the second difference value is not smaller than beta, early warning is carried out; beta is a departure time threshold;

And if the second difference value is smaller than beta, before the marking is finished, determining that the personnel appearing leave the marking chamber according to the tracking result, storing the appearance time and the leaving time by taking the identification of the personnel appearing as the identification, and updating beta according to the leaving times.

Optionally, updating β according to the number of exits includes:

acquiring the departure time stored in each departure;

acquiring the occurrence time stored when the user leaves for the first time;

Calculating each departure time = occurrence time stored at a last departure-departure time stored at a previous departure;

Calculating standard deviation of each departure time;

And determining beta as a minimum threshold value [ 1+standard deviation.

Optionally, after the identity confirmation is passed, the method further comprises:

identifying the physical actions of the people according to the image tracking;

and if the recognized body motion has abnormal motion, early warning is carried out.

Optionally, after S101, the method further includes:

Determining whether the image is an image of a comment room;

And if the image of the label evaluation room is not evaluated, early warning is carried out.

Optionally, the method further comprises:

Acquiring audio of a label evaluation room in real time;

carrying out semantic analysis on the audio;

And if the semantic analysis result comprises abnormal semantics, early warning is carried out.

Optionally, after acquiring the audio of the label evaluation room in real time, the method further comprises:

identifying sounds in the audio;

if no sound is found, early warning is carried out.

(III) beneficial effects

Acquiring an image of a bid evaluation room in real time between the beginning and the end of bid evaluation; based on the image, each time it is determined that personnel are present at the entrance of the rating room: recording the appearance time and tracking the appearance personnel; determining whether a human face appears according to the tracking result; if the face is still not recognized when the appearance time +alpha is reached, the identity confirmation is not passed; if the face is recognized within the occurrence time +alpha, the face recognition is carried out on the recognized face, if the face recognition result is that the person appearing is a label evaluation expert, the identity confirmation is passed, and if the face recognition result is that the person appearing is not a label evaluation expert, the identity confirmation is not passed; if the identity is not confirmed, early warning is carried out; if the identity is confirmed to pass, the personnel appearing are tracked in real time according to the image, and early warning is carried out according to the identification of the personnel appearing and the tracking result, so that the automatic early warning of the label evaluation room is realized.

Drawings

Fig. 1 is a schematic flow chart of an image-based marking room early warning method according to an embodiment of the invention.

Detailed Description

The invention will be better explained by the following detailed description of the embodiments with reference to the drawings.

The method is very important for early warning of abnormal conditions (such as intrusion of unauthorized personnel, abnormal equipment and the like) in the evaluation process. The existing scheme carries out early warning through a mode of manually checking the monitoring video, and the mode not only consumes labor, but also can have the condition of abnormal missing report caused by personnel errors.

Based on the above, the invention provides an image-based bid evaluation room early warning method, which comprises the following steps: acquiring an image of a bid evaluation room in real time between the beginning and the end of bid evaluation; based on the image, each time it is determined that personnel are present at the entrance of the rating room: recording the appearance time and tracking the appearance personnel; determining whether a human face appears according to the tracking result; if the face is still not recognized when the appearance time +alpha is reached, the identity confirmation is not passed; if the face is recognized within the occurrence time +alpha, the face recognition is carried out on the recognized face, if the face recognition result is that the person appearing is a label evaluation expert, the identity confirmation is passed, and if the face recognition result is that the person appearing is not a label evaluation expert, the identity confirmation is not passed; if the identity is not confirmed, early warning is carried out; if the identity is confirmed to pass, the personnel appearing are tracked in real time according to the image, and early warning is carried out according to the identification of the personnel appearing and the tracking result, so that the automatic early warning of the label evaluation room is realized.

Referring to fig. 1, the implementation process of the image-based bid evaluation room early warning method provided in this embodiment is as follows:

s101, acquiring images of the bid evaluation room in real time between the beginning and the end of bid evaluation.

Before step S101 is performed, an image capturing device (such as a camera) is installed in the label evaluation room, and there may be multiple image capturing devices or one image capturing device, and it is required that the image capturing device can capture the images in the inlet and the outlet (possibly one port or two ports) of the label evaluation room and the space of the label evaluation room.

For example, an image acquisition device is installed, and the image acquired by the device comprises images of the entrance and the exit of the marking room and images in the marking room space.

For another example, two image acquisition devices are installed, wherein the image acquired by one image acquisition device comprises the entrance image and the exit image of the evaluation room, and the image acquired by the other image acquisition device comprises the image in the space of the evaluation room. Or the image collected by one image collecting device comprises the image in the marking room entrance and marking room space, and the image collected by the other image collecting device comprises the marking room exit image. Or the image collected by one image collection device comprises the label evaluation room outlet and the image in the label evaluation room space, and the image collected by the other image collection device comprises the label evaluation room inlet image.

For another example, three image acquisition devices are installed, wherein the image acquired by one image acquisition device comprises an image of the entrance of the evaluation room, the image acquired by the other image acquisition device comprises an image of the exit of the evaluation room, and the image acquired by the other image acquisition device comprises an image in the space of the evaluation room.

For the condition of a plurality of image acquisition devices, all the image acquisition devices simultaneously acquire images and simultaneously end the image acquisition, so that each frame of image acquired by all the image acquisition devices corresponds to the image at the same moment, the comparability among the acquired images is ensured, and the time synchronism of the subsequent image processing is also ensured.

The later image of a certain time in this embodiment includes the images of the entrance, the exit, and the evaluation room of the time evaluation room.

It should be noted that, the method provided in this embodiment works in the time period when the label evaluation room performs the label evaluation, that is, the image of the label evaluation room in S101 is only the image between the beginning and the end of the label evaluation, when the label evaluation end time is reached, the execution of the method shown in fig. 1 is stopped, and when the next label evaluation is waited, the method shown in fig. 1 is re-executed.

S102, based on the image, early warning is carried out according to the content of the image every time when personnel appear at the entrance of the evaluation room.

In a specific implementation, when S102 is executed, the following steps are executed each time a person is determined to be present at the entrance of the evaluation room based on the image:

For example, after determining that the personnel i appear at the entrance of the bid evaluation room based on the image, the personnel i may be a bid evaluation expert, and the bid evaluation expert enters the bid evaluation room for the first time; the person i may be a label evaluation expert, but the label evaluation expert enters the label evaluation room once in the label evaluation process, leaves the label evaluation room subsequently, and enters the label evaluation room again; person i can also be other persons than the label expert. After personnel i appear at the entrance of the calibration room, the following process is executed:

s102-1, determining a pre-allocation identifier, and setting the identifier of the personnel to be appeared as the pre-allocation identifier.

The preassigned identifier here is an identifier assigned to the person present and not the final identifier of that person. The pre-allocation identifier can only indicate how many times the person enters the rating room, and does not distinguish whether the person is the same person. That is, for person i, if it enters the rating room for the first time, it is determined that the pre-allocation flag is 1, then it leaves the rating room, then enters the rating room again, and if no other person enters the rating room during the period from the first time to the time of entering the rating room again, then it is determined that its pre-allocation flag is 2 when entering the rating room again. For another example, for person i, if it enters the rating room for the first time, it determines its pre-allocation flag as 1, then it leaves the rating room, then enters the rating room again, if only person j enters the rating room during the period from the first time to the second time, then it determines its pre-allocation flag as 3 when it enters the rating room again, and it determines its pre-allocation flag as 2 when it enters the rating room.

S102-2, recording the occurrence time and tracking the people who appear.

The appearance time is the time when people appear at the entrance of the evaluation room door are first identified through the object identification technology in the existing image.

It should be noted that, the "first time" herein is not the absolute first time, that is, it is not the first person who enters the label evaluation room, but the entrance of the label evaluation room of the previous frame is not personnel, the entrance of the label evaluation room of the next frame is personnel, that is, after the entrance of the label evaluation room is not personnel, personnel appear, and at this time, the time of the next frame is the appearance time. Or the entrance of the evaluation room of the previous frame has x persons, the entrance of the evaluation room of the next frame has x+1 persons, namely, the entrance of the evaluation room has a new person, and the time of the next frame is the appearance time.

Since people appear at the entrance each time, the people are tracked by the object tracking technology in the existing image, when untracked people appear in the current image frame, new people can be considered to appear.

S102-3, determining whether a human face appears according to the tracking result.

In step S102-2, only entry of a person is recognized, but not necessarily a face. The panelist needs to check in after entering the label-evaluating room, one check-in mode is to face identification for the image acquisition equipment to finish check-in, and the proposal aims at the situation, so that after the fact that new personnel enter is determined, each frame of subsequent images can be continuously identified, and whether the tracked personnel have a face or not is determined.

The prior art used in the technique of determining whether a face is present in an image frame is not described in any great detail herein.

S102-4, if the face is still not recognized when the appearance time +alpha is reached, the identity confirmation is not passed. If the face is recognized within the occurrence time +alpha, the face recognition is carried out on the recognized face, if the face recognition result is that the person appearing is the evaluation expert, the identity confirmation is passed, and if the face recognition result is that the person appearing is not the evaluation expert, the identity confirmation is not passed.

And alpha is a preset time threshold, is a preset standard value, if written in an operation guide of a label evaluation expert, enters a label evaluation room to finish face recognition within 3 minutes, and signs in, then alpha=3 minutes.

If the appearance time +alpha is reached, the face is still not recognized, and the person i is not executed according to the evaluation standard, and the identity confirmation is determined to be failed.

If the face is recognized within the occurrence time +alpha, the face recognition is carried out on the recognized face, if the face recognition result is that the person appearing is the evaluation expert, the identity confirmation is passed, and if the face recognition result is that the person appearing is not the evaluation expert, the unauthorized user is indicated to enter the evaluation room, and the identity confirmation is not passed.

S102-5, if the identity confirmation is not passed, the sign-in process is abnormal, or an unauthorized user enters the label evaluation room to cause the label evaluation process to be abnormal, so that early warning is carried out, no subsequent steps are carried out after early warning, and step S102 is repeatedly carried out (that is, after the early warning is carried out on the current personnel, the current personnel is not tracked, and whether a new personnel enters an entrance is determined again). In addition, the early warning scheme can be used for making a call to related personnel, playing a corresponding warning sound for abnormal behaviors and the like.

If the identity confirmation is passed, the personnel i is an authorized user, and the behavior accords with the standard of evaluation, and the following steps are performed:

1. And updating the identification of the personnel to be appeared as the identification of the evaluation expert.

Because the face recognition is performed, the identity of the person i is known at the moment, and the identity of the person i is updated to the identity of the expert, so that if the newly entered person is an authorized person with standard behaviors (namely, the label evaluation expert), the final identity is the identity of the label evaluation expert; if the newly entered person is an unauthorized person or the behavior is not specified, then the identification is simply a pre-assigned identification.

2. And tracking the people in real time according to the images, and carrying out early warning according to the identification and tracking results of the people.

In particular, the method comprises the steps of,

1) And determining whether the personnel enter the marking room for the first time according to the identification of the personnel.

Because each time the person passing the identity confirmation leaves the label evaluation room, the person identification is taken as the identification to store the leaving time, the step can inquire the stored leaving time, and if the stored leaving time identification has the identification of the person i, the person is considered to enter the label evaluation room for the first time. If the stored identifier of the departure time does not identify the person i, the person i is considered to enter the rating room for the first time.

2) If the label evaluation room is accessed for the first time, the following steps are carried out:

1.1 before the end of the bid evaluation, determining whether the personnel leaves the bid evaluation room according to the tracking result.

And 1.2, if the personnel appearing are determined not to leave the marking room according to the tracking result after marking is finished, the marking process standard is described, no early warning is carried out, and the image-based marking room early warning method provided by the embodiment is stopped to be executed after the marking finishing time is reached.

1.3, If the person who appears is determined to leave the rating room according to the tracking result before the rating is finished, because the person enters the rating room for the first time, the person also leaves the rating room for the first time at the moment, then:

1.3.1 storing departure times with the identity of the person present as the identity.

For example, for person i, identified as IDi, the time of this departure is XX year XX month XX day xx.xx.xx, then XXXX year XX month XX day xx.xx.xx will be stored, and the identification of this XX year XX month XX day xx.xx.xx value is IDi.

1.3.2 Calculating a departure time period threshold value beta and storing beta.

The calculation process of the departure time length threshold value beta is as follows:

(1) And acquiring whether the personnel present submit the bid evaluation result.

(2) If the personnel present have submitted the evaluation result, the personnel i is explained that the evaluation work is completed, and then:

a. first difference = rating end time-departure time is calculated.

The first difference characterizes the time that person i leaves the rating room from the end of the rating.

B. And when the first difference value is not greater than the preset minimum threshold value, determining beta as a preset maximum value.

The minimum threshold is a preset standard value, for example, 5 minutes before the end of the bid evaluation, and the bid evaluation expert who completes the bid evaluation can leave the bid evaluation room, so the minimum threshold is 5 minutes.

When the first difference is not greater than the preset minimum threshold, it indicates that the person i leaves the evaluation room to perform normally, and then β is a preset maximum value, and further in step 1.3.3, the time of leaving time +β is also a maximum value, so that the evaluation is finished before the time of reaching leaving time +β, and further the execution of the method shown in fig. 1 is stopped, and no early warning condition occurs.

It should be noted that, the method provided in this embodiment works in the time period when the label evaluation room performs the label evaluation, that is, the image of the label evaluation room in S101 is only the image between the beginning and the end of the label evaluation, when the label evaluation end time is reached, the execution of the method shown in fig. 1 is stopped, and when the next label evaluation is waited, the method shown in fig. 1 is re-executed.

C. When the first difference is greater than a preset minimum threshold, β is determined as the minimum threshold [1+ (total time length of rating-first difference)/total time length of rating ].

When the first difference is greater than the preset minimum threshold, the explanatory person i is in violation of the standard behavior, who leaves the rating room during the rating. However, the rating specialist may leave the rating room with various reasonable demands (such as a bathroom), and therefore, a minimum threshold value, which is an empirical value, is preset, mainly for the purpose of normalizing the time of normal demands. For example, the sample is analyzed to obtain a normal toilet-going time of 4 minutes, and then the minimum threshold is 4 minutes.

In addition, the premise of entering c is that the personnel who appears has submitted the bid evaluation result, that is, the personnel i has completed the bid evaluation work, at this time, the departure of the bid evaluation expert is in violation of the standard, but the stay time of the personnel i in the bid evaluation room is not required to meet the standard because of the bid evaluation work problem, for the situation, the bid evaluation expert may slow down because the personnel i completes the bid evaluation work and act as exceeding the average value, so that the time length is enlarged again on the basis of 4 minutes for the personnel i, that is, the method provided by the embodiment does not judge according to the minimum threshold value, but adds a certain time reservation on the minimum threshold value, if the bid evaluation room is returned during the time reservation, the early warning is not performed, and if the bid evaluation room is not returned after the time reservation is added, the early warning is performed.

The time reservation is dynamically determined according to the situation that the person i has continuously rated the time length (i.e. the total rated time length-the first difference value), the longer the time length has been continuously rated (i.e. the greater the value of the total rated time length-the first difference value)/the longer the total rated time length is, the greater the demand for reasonably leaving the rated room is, and the slow degree of subconscious action after leaving is increased, so the time reservation is slightly more, and the time reservation is the time with the minimum threshold value (the total rated time length-the first difference value)/the total rated time length, that is, the time reservation is the time with the maximum minimum threshold value. For the case that the time reservation is a minimum threshold time, it also indicates that the evaluation is finished, and then the execution of the method shown in fig. 1 is stopped, so that no early warning condition occurs.

Thus, β is determined as the minimum threshold + time reservation (i.e., time reservation is the minimum threshold x (total bid-length-first difference)/total bid-length) =minimum threshold x [1+ (total bid-length-first difference)/total bid-length ].

(3) If the personnel do not submit the evaluation result, the method comprises the following steps:

i. When the first difference value is not greater than a preset minimum threshold value, the fact that the personnel i does not complete the label evaluation work still leaves the label evaluation room is explained, the remaining effective label evaluation time is very short, early warning is needed in time, and therefore beta is determined to be 0.

And ii, when the first difference value is larger than a preset minimum threshold value, the fact that the personnel i does not finish the bid evaluation work and still leaves the bid evaluation room is explained, but the remaining effective bid evaluation time can meet the normal leave requirement of the personnel i, so that beta is determined to be the minimum threshold value.

It should be noted that, each label-evaluating expert corresponds to a β, and after calculating β, the β may be identified by using the identifier of the person i as the identifier.

The normal leaving requirements of all bid evaluation experts are met according to the real bid evaluation conditions through dynamic calculation beta, and meanwhile, abnormal leaving behaviors of a bid evaluation room are identified.

1.3.3 If the departure time is +beta, the personnel do not enter the evaluation room again, and early warning is carried out.

That is, if the leaving time exceeds β, the warning is performed, that is, the person's behavior of leaving the rating room is abnormal.

In addition, no subsequent steps are performed after the early warning, and step S102 is repeatedly performed (that is, after the current person performs the early warning, the current person is not tracked any more, and whether a new person enters the entrance is newly determined). In addition, the early warning scheme can be used for making a call to related personnel, playing a corresponding warning sound for abnormal behaviors and the like.

In addition, there may be a case where the departure time+β is longer than the evaluation end time, and since the execution of the method shown in fig. 1 is automatically ended at the evaluation end time, the execution of the method shown in fig. 1 is actually ended with the end of the execution of the method shown in fig. 1 for a case where the departure time+β is longer than the evaluation end time, and no warning is performed.

3) If the label is not firstly entered into the label evaluation room, the following steps are carried out:

2.1 obtaining the last departure time according to the identification of the personnel.

Because each time the person passing the identity confirmation leaves the label evaluation room, the person identification is taken as the identification to store the leaving time, the step can inquire all the leaving times corresponding to the identification of the person appearing, and the leaving time closest to the current time is taken as the last leaving time.

2.2 Calculate the second difference = time of occurrence-last time of departure.

The second difference characterizes the length of time that the person i enters the room from the last departure, i.e. the last departure from the room.

And 2.3, if the second difference value is not smaller than beta, early warning is carried out.

Where β is the departure time threshold.

When the person i leaves the label evaluation room for the first time, beta is calculated (the calculating process is shown in step 1.3.2), so that when the person i leaves the label evaluation room again, the beta calculated in the last time is directly read.

Beta represents the longest time that the personnel i normally leaves the evaluation room, and if the second difference value is not smaller than beta, the personnel i leaves overtime, so that early warning is carried out. Meanwhile, no subsequent step is performed after the early warning, and step S102 is repeatedly performed (that is, after the current person performs the early warning, the current person is not tracked any more, and whether a new person enters the entrance is newly determined). In addition, the early warning scheme can be used for making a call to related personnel, playing a corresponding warning sound for abnormal behaviors and the like.

In addition, each comment expert corresponds to one beta, and after the beta is calculated, the beta is identified by taking the identification of the person i as the identification. Thus, this step will obtain the corresponding β by the identity of the person present.

In addition, when leaving the label evaluation room for the first time, the step (i.e. step 1.3.3) is also performed based on beta, the step is also performed based on beta, the two pre-warning is not repeated, but different pre-warning is performed, and the pre-warning purpose in step 1.3.3 is to perform pre-warning on the personnel with abnormal leaving so as to prompt the personnel in the label evaluation room to have abnormal leaving. The early warning purpose of this step is that the personnel that unusual entering carries out the early warning to the personnel appear unusual appearance in the suggestion evaluation room. If the same person leaves the device for the same time, and the abnormal leaving and the abnormal entering are met, the device can perform early warning for two times. If a person does not enter after leaving, abnormal leaving early warning in step 1.3.3 is performed.

2.4 If the second difference value is smaller than beta, the last leaving time is normal, so that tracking is continued, when the person appearing is determined to leave the evaluation room according to the tracking result before the evaluation is finished, the appearing time and the leaving time are stored by taking the label of the person appearing as the label, and beta is updated according to the leaving times.

That is, the departure time is stored with only the identity of the person present at the first departure, and the other person present at each departure is stored with the identity of the person present as the identity.

In addition, the step 1.3.2 is that the threshold value β of the leaving time period is calculated when the label evaluation expert leaves the label evaluation room for the first time, and then the threshold value β of the leaving time period is updated every time the label evaluation room leaves, and the updating process is as follows:

S201, acquiring the departure time stored in each departure.

Because the departure time is stored by taking the identity of the person appearing as the identity of the person who leaves for the first time or not, the step searches all the departure times stored by the identity according to the identity of the person appearing.

After the search, all the departure times are sorted according to the distance between the departure time and the current time, and then the departure times are sorted into a departure time sequence from far to near, for example, the departure time sequence is { departure time 1, departure time 2, departure time 3}. The first element in the departure time sequence is the first departure time, the second element is the second departure time, the third element is the third departure time, etc.

S202, acquiring the occurrence time stored when the user leaves for the first time.

Since the first departure only uses the identification of the person appearing as the identification to store the departure time, the other people who appear at each departure use the identification of the person appearing as the identification to store the appearance time and the departure time. Thus, the time of occurrence is stored only when it is not first left. This step will therefore look up all times of occurrence stored by the identity of the person present.

After the search, all the appearance times are ranked according to the distance between the appearance time and the current time, and then the appearance times are ranked into an appearance time sequence from far to near, if the appearance time sequence is { appearance time 1, appearance time 2}.

For example, after the first entry, person i leaves the rating mark at leaving time 1, at which point leaving time 1 is recorded. After this, person i enters the rating room a second time at appearance time 1 and leaves the rating room at departure time 2, at which time two times are recorded at the second departure, appearance time 1 and departure time 2. Then person i enters the rating room a second time at appearance time 2 and leaves the rating room at departure time 3, at which time two times, appearance time 2 and departure time 3, are recorded at the third departure. … … A

It can be seen that the 1 st element in the departure time sequence is the first departure time, and the 1 st element in the appearance sequence is the second entry time recorded at the second departure time. The x-th element in the appearance time sequence is the x+1-th entry time of the ground recorded at the x+1-th exit, and the x-th element in the exit time sequence is the x-th exit time.

S203, calculate each departure time=the occurrence time stored at the time of the last departure-the departure time stored at the time of the previous departure.

Because the 1 st element in the departure time sequence is the first departure time, the 1 st element in the appearance sequence is the second entry time recorded at the second departure. The x-th element in the appearance time sequence is the x+1-th entry time of the ground recorded at the x+1-th exit, and the x-th element in the exit time sequence is the x-th exit time. The first departure time is thus the occurrence time stored at the second departure (i.e. the first element in the occurrence sequence, i.e. occurrence time 1) -the departure time stored at the first departure (i.e. the first in the departure sequence, i.e. departure time 1). Accordingly, the xth departure time is the occurrence time stored at the xth+1th departure (i.e., the xth element in the appearance sequence, i.e., the occurrence time x) -the departure time stored at the xth departure (i.e., the xth element in the departure sequence, i.e., the departure time x).

S204, calculating standard deviation of the departure time of each two adjacent times.

The labeling differences are calculated by the existing standard deviation calculation scheme, and are not described in detail herein.

The standard deviation characterizes the degree of difference of the departure time of each departure, and the larger the standard deviation is, the larger the difference of each departure is, and the smaller the regularity of the duration of each departure is, the larger the randomness is; the smaller the standard deviation is, the smaller the difference of each departure is, the larger the regularity of the duration of each departure is, and the smaller the randomness is.

S205, determining β as the minimum threshold [1+ standard deviation.

Wherein the minimum threshold is the minimum threshold in step 1.3.2, c. In this step, time reservation is also performed on the basis of the minimum threshold, but the time reservation calculation scheme at this time is different from the time reservation calculation scheme at the first departure time, and the time reservation is based on the departure times and whether there is a rule.

The smaller the standard deviation is a fraction, the more regular each departure is described, then each departure may have the same purpose, the departure is not allowed in the theoretical evaluation process, and the generation of normal demands is random, therefore, the larger the regularity is, the more abnormal the description is, the smaller the reserved time is, and when the departure times are the same, the smaller the standard deviation is, the smaller the value of the number of the standard deviation is, and the smaller the value of the time reservation (i.e. the minimum threshold value is (the number of the standard deviation is)) is.

The more and the more abnormal the number of leaves, the smaller the reservation time, and when the standard deviation is the same, the larger the number of leaves, the smaller the value of the standard deviation ≡the number of leaves, and the smaller the value of the time reservation (i.e. minimum threshold ≡the number of leaves).

Therefore, the calculation of the time reservation (i.e. the minimum threshold value (standard deviation separation frequency)) in this embodiment is considered from two dimensions of separation frequency and regularity of separation, which better meets the reasonable requirements of users.

Β is minimum threshold + time reserved = minimum threshold [1+ standard deviation.

The method is characterized in that an early warning scheme of how an unauthorized person enters the bid evaluation room and a bid evaluation expert leaves another scene of the bid evaluation room is adopted, and in addition, the abnormal action of the bid evaluation expert can be early warned after the identity confirmation is passed.

For example: after the identity confirmation is passed, the physical actions of the person are recognized according to the image tracking. And if the recognized body motion has abnormal motion, early warning is carried out.

The existing recognition scheme is adopted for the recognition of the body motion, such as determining the skeleton points of people appearing in each frame of image, and determining the behaviors of the people appearing according to the relation between the skeleton points.

In addition, the abnormal actions are preset and flexibly determined according to specific situations. For example, the evaluation mark is that the evaluation mark expert cannot enter the evaluation mark room with the mobile phone, the abnormal action can be set as the action of making a call, and if the person makes the call according to the relation between the bone points, the alarm is given. There may be a plurality of abnormal actions.

In addition, the method of the present embodiment also identifies the image acquired in step S101, that is, determines whether the image is an image of the comment room after executing step S101. If the image of the non-standard evaluation room (such as a black screen caused by the shielding of the image acquisition equipment, or a black screen caused by the failure of the image acquisition equipment, or a black screen caused by other factors, or the image of the non-standard evaluation room caused by other factors) is subjected to early warning.

The image acquisition device can be tested between the evaluation marks, so that a correct evaluation mark room image (a standard point can be set in the evaluation mark room, for example, a special plant is used as the standard point) acquired during the test is reserved as a standard image, after each frame of image is acquired in the step S101, the acquired image and the standard image are not matched (for example, the position of the standard point is compared), if the acquired image and the standard image are matched, the image is considered as the evaluation mark room image, otherwise, the image is considered as abnormal, and an alarm is given.

In addition, the audio in the bid evaluation room can be acquired through the audio acquisition equipment, and the audio in the bid evaluation room is subjected to early warning, namely the audio in the bid evaluation room is acquired in real time. Semantic analysis is performed on the audio. And if the semantic analysis result comprises abnormal semantics, early warning is carried out.

The configuration of the audio collection equipment can configure one or more pieces of equipment according to actual conditions, so that all the audio in the comment room can be clearly collected. Existing speech analysis schemes employed for speech analysis are used to determine what the sound in the rating room characterizes. The abnormal semantics are also preset, and can be one semantics or a plurality of semantics according to specific situations, for example, the semantics of the bid winning of the hinting bidder are taken as the abnormal semantics.

In addition, sounds in the audio are also identified. If no sound is found, the audio acquisition equipment is abnormal, and early warning is carried out.

In practical application, there is a possibility that there is no sound in a short time, so that no sound can be found in a continuous preset time period (for example, 3 minutes), and then early warning is performed.

The duration here may be empirically set.

The method provided by the embodiment acquires the image of the evaluation room in real time between the start and the end of the evaluation; based on the image, each time it is determined that personnel are present at the entrance of the rating room: recording the appearance time and tracking the appearance personnel; determining whether a human face appears according to the tracking result; if the face is still not recognized when the appearance time +alpha is reached, the identity confirmation is not passed; if the face is recognized within the occurrence time +alpha, the face recognition is carried out on the recognized face, if the face recognition result is that the person appearing is a label evaluation expert, the identity confirmation is passed, and if the face recognition result is that the person appearing is not a label evaluation expert, the identity confirmation is not passed; if the identity is not confirmed, early warning is carried out; if the identity is confirmed to pass, the personnel appearing are tracked in real time according to the image, and early warning is carried out according to the identification of the personnel appearing and the tracking result, so that the automatic early warning of the label evaluation room is realized.

In order that the above-described aspects may be better understood, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions.

It should be noted that 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.

Furthermore, it should be noted that in the description of the present specification, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to a specific feature, structure, material, or characteristic described in connection with the embodiment or example being included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example.

While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art upon learning the basic inventive concepts. Therefore, the appended claims should be construed to include preferred embodiments and all such variations and modifications as fall within the scope of the invention.

It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, the present invention should also include such modifications and variations provided that they come within the scope of the following claims and their equivalents.

Claims (8)

1. An image-based bid evaluation room early warning method is characterized by comprising the following steps:

s101, acquiring an image of a bid evaluation room in real time between the beginning and the end of bid evaluation;

S102, based on the image, executing the following steps after each time that personnel appear at the entrance of the evaluation room is determined:

Determining a pre-allocation identifier, and setting the identifier of the person appearing as the pre-allocation identifier;

recording the occurrence time and tracking the people who appear;

determining whether a human face appears according to the tracking result;

If the face is still not recognized when the appearance time +alpha is reached, the identity confirmation is not passed; if the face is recognized within the occurrence time +alpha, the face recognition is carried out on the recognized face, if the face recognition result is that the person appearing is a label evaluation expert, the identity confirmation is passed, and if the face recognition result is that the person appearing is not a label evaluation expert, the identity confirmation is not passed; wherein alpha is a preset duration threshold;

If the identity is not confirmed, early warning is carried out; if the identity is confirmed to pass, updating the identification of the personnel appearing as the identification of the rating expert, tracking the personnel appearing in real time according to the image, and carrying out early warning according to the identification of the personnel appearing and the tracking result;

and carrying out early warning according to the identification and tracking results of the people, wherein the early warning comprises the following steps:

Determining whether the personnel enter the evaluation room for the first time according to the identification of the personnel;

if the person enters the evaluation room for the first time, before the evaluation is finished, determining that the person leaves the evaluation room according to the tracking result, storing the leaving time by taking the mark of the person as the mark, calculating a leaving time threshold beta, and storing the beta; if the departure time is +beta, the personnel does not enter the evaluation room again, and early warning is carried out;

The calculating the departure time threshold value beta comprises the following steps:

Acquiring whether the personnel present submit a bid evaluation result;

if the personnel present have submitted the bid evaluation result, calculating a first difference value = bid evaluation ending time-the departure time; when the first difference value is not greater than a preset minimum threshold value, determining beta as a preset maximum value; and when the first difference value is larger than a preset minimum threshold value, determining beta as the minimum threshold value [1+ (total evaluation duration-first difference value)/total evaluation duration ].

2. The method of claim 1, wherein after the step of obtaining whether the person present submitted the bid evaluation result, further comprises:

if the personnel do not submit the evaluation result, determining beta as 0 when the first difference value is not larger than a preset minimum threshold value; and when the first difference value is larger than a preset minimum threshold value, determining beta as the minimum threshold value.

3. The method of claim 1, wherein said determining if said person is first entering said assessment room further comprises:

if the person enters the evaluation room for the first time, acquiring the last leaving time according to the mark of the person;

calculating a second difference = said time of occurrence-last time of departure;

if the second difference value is not smaller than beta, early warning is carried out; the beta is a departure time threshold;

And if the second difference value is smaller than beta, before the marking is finished, determining that the personnel appearing leave the marking chamber according to the tracking result, storing the appearance time and the leaving time by taking the identification of the personnel appearing as the identification, and updating beta according to the leaving times.

4. A method according to claim 3, wherein updating β according to the number of exits comprises:

acquiring the departure time stored in each departure;

acquiring the occurrence time stored when the user leaves for the first time;

Calculating each departure time = occurrence time stored at a last departure-departure time stored at a previous departure;

Calculating standard deviation of each departure time;

and determining beta as the minimum threshold value [ 1+standard deviation.

5. The method of claim 1, further comprising, after the identity confirmation passes:

identifying a physical action of the person present from the image tracking;

and if the recognized body motion has abnormal motion, early warning is carried out.

6. The method according to claim 1, further comprising, after S101:

determining whether the image is an image of the comment room;

and if the image of the evaluation room is not the image, early warning is carried out.

7. The method according to claim 1, wherein the method further comprises:

Acquiring audio of a label evaluation room in real time;

semantic analysis is carried out on the audio;

And if the semantic analysis result comprises abnormal semantics, early warning is carried out.

8. The method of claim 7, wherein after the audio of the rating room is obtained in real time, further comprising:

identifying sounds in the audio;

if no sound is found, early warning is carried out.