Clinical test project man-hour analysis management system

Abstract

The invention belongs to the field of man-hour analysis and management, and discloses a clinical test project man-hour analysis and management system, which comprises a man-hour analysis and management device, and is characterized in that the man-hour analysis and management device comprises a video analysis module, and the video analysis module comprises an acquisition unit, an analysis unit and a statistical unit; the acquisition unit is used for storing all frames contained in a monitoring video into a set FraSet according to the sequence of shooting time from morning to evening; the analysis unit is used for respectively judging whether each frame in the FraSet contains an attendance employee or not; according to the invention, when the attendance staff exists in the currently identified frame, the total number of identification areas needing to be identified for identifying the attendance staff in the current identified frame can be effectively reduced, and the identification speed is improved. Therefore, the total time consumption for obtaining the identification results of all the frames of the monitoring video can be shortened, and the efficiency of judging the effectiveness of working hours is improved.

Description

Technical Field

[0001] The present invention relates to the field of work time analysis management, and in particular to a work time analysis management system for clinical trial projects. Background Art

[0002] Working hours management in clinical trials is an important method to ensure that clinical trials can proceed smoothly according to the established plan. In the process of management, working hours can be counted by automatically calculating working hours or by employees manually reporting working hours. In order to improve the effectiveness of working hours, cameras are usually set up in clinical trial sites to capture videos of employees at work, and then the videos are analyzed to determine the effectiveness of working hours. In the analysis process, the judgment is generally made by calculating the ratio between the total length of time employees stay in the clinical trial site and the working hours reported by the employees or the working hours calculated based on the clocking in and out time. When the ratio is higher than the set threshold, it means that the credibility of the working hours on that day is high.

[0003] In the prior art, each frame in the video is usually detected to determine whether the employee is at the work station in the frame. However, the duration of the video of the employee at work is generally long and contains a large number of frames, which results in a long time to obtain the recognition result of each frame. Summary of the invention

[0004] The purpose of the present invention is to disclose a clinical trial project labor time analysis management system to solve the problem of how to improve the efficiency of judgment when using video to judge the effectiveness of labor time in the process of labor time analysis management of clinical trial projects.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] The present invention provides a clinical trial project work time analysis management system, including a work time analysis management device, the work time analysis management device includes a video analysis module, the video analysis module includes an acquisition unit, an analysis unit and a statistical unit;

[0007] The acquisition unit is used to save all frames contained in the surveillance video into a set FraSet in the order of shooting time from morning to night;

[0008] The analysis unit is used to determine whether each frame in FraSet contains employees who are present, including:

[0009] The process of identifying the first and second frames is as follows:

[0010] The second detection algorithm is used to identify the hth frame to determine whether there is an employee on duty in the first frame;

[0011] The process of identifying the hth frame is as follows:

[0012] h is a positive integer greater than or equal to 3, and it is determined whether there are employees present in both the h-1th frame and the h-2th frame. If so, the first detection algorithm is used to identify the hth frame to determine whether there are employees present in the hth frame; if not, the second detection algorithm is used to identify the hth frame to determine whether there are employees present in the hth frame;

[0013] The first detection algorithm includes:

[0014] The first step is to calculate the size of the recognition area based on the h-1th frame and the h-2th frame;

[0015] The second step is to calculate the detection value of the pixel point in the hth frame based on the h-1th frame and the h-2th frame;

[0016] The third step is to save all the pixels in the hth frame into the set PixSet in descending order of detection values. h ;

[0017] Step 4: Initialize the value of k to 1;

[0018] Step 5: PixSet h The kth pixel point in is taken as the center of the recognition area, and the recognition area is identified to determine whether there is an employee on duty in the recognition area. If yes, the detection is stopped, otherwise, the sixth step is entered;

[0019] Step 6: Update the value of k, add 1 to the value of k, and go to step 7;

[0020] Step 7: Determine whether the value of k is greater than PixSet h The total number of pixels in , if yes, it means that there is no employee present in the hth frame, if no, go to step 5;

[0021] The statistical unit is used to count the total time that employees are in the clinical trial site in the surveillance video.

[0022] Preferably, it also includes a working time acquisition device, which is used to acquire the working hours of employees of the clinical trial project on the attendance day.

[0023] Preferably, the working time analysis and management device further includes a judgment module and a working time management module;

[0024] The judgment module is used to compare the total time with the working hours to determine whether the working hours are abnormal;

[0025] The working hours management module is used by working hours management personnel to correct abnormal working hours and obtain corrected working hours.

[0026] Preferably, it also includes a video shooting device, which is used to obtain monitoring videos of the clinical trial site of the employees of the clinical trial project on the attendance day.

[0027] Preferably, the second detection algorithm comprises:

[0028] Taking each pixel point in the hth frame as the center of a recognition area of ​​a preset size, a plurality of recognition areas are obtained;

[0029] Each identification area is identified respectively to determine whether there are employees on duty in all the identification areas. Preferably, when identifying the identification areas, an R-CNN algorithm or a YOLO algorithm is used to identify the identification areas.

[0030] Preferably, calculating the size of the recognition area based on the h-1th frame and the h-2th frame includes:

[0031] Use (x h-1 ,y h-1 ) represents the coordinates of the center of the recognition area with the most pixels in the area containing employees on duty in the h-1th frame;

[0032] Use (x h-2 ,y h-2 ) represents the coordinates of the center of the recognition area containing the most pixels belonging to employees on duty in the h-2th frame;

[0033] In the hth frame, make a h-1 ,y h-1 ) and (x h-2 ,y h-2 ) of the straight line L h-1 ;

[0034] Calculate (x h-1 ,y h-1 ) and (x h-2 ,y h-2 ) Distance D h-1,h-2 ;

[0035] Get except (x h-2 ,y h-2 ), and (x h-1 ,y h-1 ) is equal to D h-1,h-2 And through L h-1 The coordinates of the point (x pre ,y pre );

[0036] Based on (x pre ,y pre )Calculate the size of the recognition area.

[0037] Preferably, based on (x pre ,y pre ) Calculate the size of the recognition area, including:

[0038] Use Lstd h-1 and Lstd h-2 represents the difference between the maximum and minimum values ​​of the ordinates of the pixel points belonging to the employee in the h-1th frame and the h-2th frame;

[0039] Use Wstd h-1 and Wstd h-2 It represents the difference between the maximum and minimum values ​​of the horizontal coordinates of the pixel points belonging to the employee in the h-1th frame and the h-2th frame;

[0040] Then in the hth frame, the calculation formula for the length of the recognition area is:

[0041]

[0042] The calculation formula for the width of the recognition area is:

[0043]

[0044] L h and W h Respectively represent the length and width of the recognition area in the hth frame; AL h and A.W. h Respectively represent the length and width of the hth frame,

[0045] Preferably, calculating the detection value of the pixel point in the h-th frame based on the h-1-th frame and the h-2-th frame includes:

[0046] Use b to represent the pixel in the hth frame, and use check b Indicates the detection value of b, then check b The calculation formula is as follows:

[0047]

[0048] dstna b Represents the coordinates of b and the coordinates (x pre ,y pre ), dstna ma Represents the pixel point and coordinate (x pre ,y pre ) is the maximum distance between b Represents b and ray FL h-1 The distance between ma Represents the pixel point in the hth frame and the ray FLh-1 The maximum value of the distance between, η1 represents the first detection coefficient, η2 represents the second detection coefficient; FL h-1 For (x h-2 ,y h-2 ) is the endpoint and passes through (x h-1 ,y h-1 ) rays.

[0049] Preferably, the total time that the employee is in the clinical trial site in the statistical monitoring video includes:

[0050] Let T1 represent the total duration of the surveillance video, N represent the total number of frames contained in the surveillance video, and M represent the total number of frames of employees present. The calculation formula for the total duration of employees in the clinical trial site is as follows:

[0051]

[0052] T2 represents the total time employees spend at the clinical trial site.

[0053] Beneficial effects:

[0054] Compared with the existing method of judging whether the frames of the surveillance video contain employees who are on duty, the present invention refers to the recognition results of the first two frames that have been judged and whose shooting time is closest to the frame currently being identified during the judgment process. When the movement of the employees on duty in the clinical trial environment is continuous, the recognition process can first identify the recognition area with a high probability of the employees on duty, so that when there are employees on duty in the frame currently being identified, the total number of recognition areas required to identify the employees on duty in the frame can be effectively reduced, thereby improving the recognition speed. In this way, the total time consumed to obtain the recognition results of all frames of the surveillance video can be shortened, and the efficiency of judging the effectiveness of working hours can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0056] Figure 1 A schematic diagram of the clinical trial project work time analysis and management system of the present invention.

[0057] Figure 2 Another schematic diagram of the clinical trial project work time analysis and management system of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] like Figure 1 As shown, the present invention provides a clinical trial project work time analysis management system, including a work time analysis management device, the work time analysis management device includes a video analysis module, and the video analysis module includes an acquisition unit, an analysis unit and a statistical unit;

[0060] The acquisition unit is used to save all frames contained in the surveillance video into a set FraSet in the order of shooting time from morning to night;

[0061] The analysis unit is used to determine whether each frame in FraSet contains employees who are present, including:

[0062] The process of identifying the first and second frames is as follows:

[0063] The second detection algorithm is used to identify the hth frame to determine whether there is an employee on duty in the first frame;

[0064] Since the two earliest frames do not have enough frames for reference, the first two frames are detected using the second detection algorithm;

[0065] The process of identifying the hth frame is as follows:

[0066] h is a positive integer greater than or equal to 3, and it is determined whether there are employees present in both the h-1th frame and the h-2th frame. If so, the first detection algorithm is used to identify the hth frame to determine whether there are employees present in the hth frame; if not, the second detection algorithm is used to identify the hth frame to determine whether there are employees present in the hth frame;

[0067] The first detection algorithm includes:

[0068] The first step is to calculate the size of the recognition area based on the h-1th frame and the h-2th frame;

[0069] The second step is to calculate the detection value of the pixel point in the hth frame based on the h-1th frame and the h-2th frame;

[0070] The third step is to save all the pixels in the hth frame into the set PixSet in descending order of detection values. h ;

[0071] Step 4: Initialize the value of k to 1;

[0072] Step 5: PixSet h The kth pixel point in is taken as the center of the recognition area, and the recognition area is identified to determine whether there is an employee on duty in the recognition area. If yes, the detection is stopped, otherwise, the sixth step is entered;

[0073] Step 6: Update the value of k, add 1 to the value of k, and go to step 7;

[0074] Step 7: Determine whether the value of k is greater than PixSet h The total number of pixels in , if yes, it means that there is no employee present in the hth frame, if no, go to step 5;

[0075] The statistical unit is used to count the total time that employees are in the clinical trial site in the surveillance video.

[0076] In the process of judgment, the present invention refers to the recognition results of the first two frames that have been judged and whose shooting time is closest to the frame currently being recognized. When the movement of the employees on duty in the clinical trial environment is continuous, the recognition process can first recognize the recognition area with a high probability of the employees on duty, so that when there are employees on duty in the frame currently being recognized, the total number of recognition areas required to recognize the employees on duty in the frame can be effectively reduced, thereby improving the recognition speed. In this way, the total time consumed to obtain the recognition results of all frames of the monitoring video can be shortened, and the efficiency of judging the effectiveness of working hours can be improved.

[0077] In addition, when performing detection, the present invention selects the detection algorithm based on whether there are employees on duty in both the h-1 frame and the h-2 frame, thereby providing more reference information for the recognition process of the h frame when recognizing the h frame, thereby improving the success rate of detection using the first detection algorithm.

[0078] When performing recognition, the present invention first calculates the detection value and arranges the detection order according to the size of the detection value. In the present invention, the larger the detection value is, the higher the detection order of the recognition area is. In this way, when there is a fish tank in the h frame, it is more possible to monitor the employees in the h frame through fewer recognition areas, thereby improving the detection efficiency of the h frame.

[0079] Preferably, if Figure 2 As shown, the clinical trial project work time analysis and management system also includes a work time acquisition device, which is used to acquire the work hours of employees of the clinical trial project on attendance days.

[0080] Preferably, the working time acquisition device includes an attendance punching device, which calculates the working hours according to the employees' clocking in and out time.

[0081] Preferably, the working time analysis and management device further includes a judgment module and a working time management module;

[0082] The judgment module is used to compare the total time with the working hours to determine whether the working hours are abnormal;

[0083] The working hours management module is used by working hours management personnel to correct abnormal working hours and obtain corrected working hours.

[0084] Preferably, the total duration is compared with the working hours to determine whether the working hours are abnormal, including:

[0085] Calculate the ratio between the total duration and the working hours, and determine whether the ratio is greater than the set threshold. If so, it means that the working hours are normal, otherwise, it means that the working hours are abnormal.

[0086] Specifically, the threshold value can be set according to actual needs. Preferably, the threshold value is 0.7.

[0087] Specifically, the abnormal working hours are corrected to obtain the corrected working hours, including:

[0088] The working time management personnel communicate with the employees to understand the reasons why the ratio is too low, and judge whether the abnormal working hours are a misjudgment based on the reasons given by the employees. If so, the abnormal individual working hours will be retained. If not, the working hours will be calculated based on the total time and working hours and corrected.

[0089] For example, if the reason why the ratio is lower than the threshold is that the employee was absent from the clinical trial site for a long time due to a temporary meeting, the abnormal working hours would be a case of misjudgment.

[0090] Preferably, the working hours are calculated based on the total time and working hours, and the working hours are corrected, including:

[0091] Calculate the average between total time and working time, and use the average as the corrected working time.

[0092] Specifically, the correction principle can be set according to actual needs. For example, the total time can be directly multiplied by a coefficient to calculate the corrected working hours. The corrected working hours are valid working hours, which is more conducive to accurately judging whether the accumulated total working hours match the standard total working hours corresponding to the progress of the clinical trial project.

[0093] Preferably, the working time management module is also used to perform the following monitoring process:

[0094] If the ratio between the total working hours consumed in the current stage and the standard working hours in this stage is less than 30%, the stage will be "green" in the progress bar of the clinical trial project, and it will be prompted that "the ratio between the total working hours consumed in the current stage and the standard working hours in this stage is less than 30%, pay attention to quality control";

[0095] If the ratio between the total working hours consumed in the current stage and the standard working hours in this stage is greater than 130%, the stage will be "red" in the progress bar of the clinical trial project, indicating "this stage is greater than the standard working hours by 130%, pay attention to improving efficiency, and control the working hours consumption in the next stage";

[0096] If the clinical trial project is progressing normally (the ratio of the total working hours consumed at this stage to the standard working hours at this stage is greater than 70% and less than 130%), this section of the progress bar will be "blue", indicating: "You have good control over your working hours, keep working hard"

[0097] Calculate the percentage of completed tasks / percentage of consumed working hours AE. If AE≈1, remind employees: the working hours should match the completion time, and attention should be paid to the balance of each link; if AE<1, remind employees: the working hours are consumed a lot, the completed tasks are few, and attention should be paid to improving efficiency; if AE>1, remind employees: the working hours are consumed little, the completed tasks are many, and attention should be paid to quality control.

[0098] Preferably, the clinical trial project work time analysis and management system further comprises a video shooting device, which is used to obtain monitoring videos of the clinical trial sites of the employees of the clinical trial project on the attendance day.

[0099] Specifically, the video shooting device is arranged at one side of the clinical trial site, and shoots the clinical trial site diagonally downward.

[0100] For example, when the clinical trial site is a rectangle, the video camera is installed at the center of the lower side of the rectangle at a height of ah. The value of ah can be set according to actual needs, for example, ah is 2.9 meters.

[0101] Preferably, the second detection algorithm comprises:

[0102] Taking each pixel point in the hth frame as the center of a recognition area of ​​a preset size, a plurality of recognition areas are obtained;

[0103] Identify each identification area separately to determine whether there are employees present in all the identification areas.

[0104] The preset size of the recognition area is the length and width of the rectangular area where the employee is located when the employee is at the center of the frame.

[0105] The second detection algorithm does not calculate the pixel detection value, and obtains multiple recognition areas by traversing the pixel points in the frame, and then recognizes the recognition areas.

[0106] The traversal order can be from top to bottom or from left to right.

[0107] Preferably, when identifying the identification area, an R-CNN algorithm or a YOLO algorithm is used to identify the identification area.

[0108] The R-CNN algorithm extracts features of the recognition area through pre-trained convolutional neural networks (such as AlexNet, VGG, etc.).

[0109] These features are input into an independent support vector machine (SVM) classifier for target classification, that is, determining whether the recognition area is a recognition area containing an employee.

[0110] The YOLO algorithm is different from traditional object detection methods. It only needs one forward propagation to predict the location and category of all objects in the image. It has obvious advantages in speed and is very suitable for real-time object detection tasks. The YOLO algorithm is trained end-to-end, which means that it can learn bounding boxes and category probabilities directly from raw images without complex pre-processing or post-processing steps. The YOLO algorithm can predict objects of different scales at the same time, which enables it to detect objects of different sizes.

[0111] Preferably, calculating the size of the recognition area based on the h-1th frame and the h-2th frame includes:

[0112] Use (x h-1 ,y h-1 ) represents the coordinates of the center of the recognition area with the most pixels in the area containing employees on duty in the h-1th frame;

[0113] Use (x h-2 ,y h-2 ) represents the coordinates of the center of the recognition area containing the most pixels belonging to employees on duty in the h-2th frame;

[0114] In the hth frame, make a pass (x h-1 ,y h-1 ) and (x h-2 ,y h-2 ) of the straight line L h-1 ;

[0115] Calculate (x h-1 ,y h-1 ) and (x h-2 ,y h-2 ) Distance D h-1,h-2 ;

[0116] Get except (x h-2 ,y h-2 ), and (x h-1 ,y h-1 ) is equal to D h-1,h-2 And through L h-1 The coordinates of the point (xpre ,y pre );

[0117] Based on (x pre ,y pre ) Calculate the size of the recognition area.

[0118] In order to obtain a larger shooting range, the video shooting device is generally set on the side of the shooting range. Therefore, the size of the employee at different positions in the frame is not the same. The prior art usually traverses all sizes to obtain the pixels actually contained in the recognition area at different sizes. This will result in a very large number of recognition areas that need to be calculated. Obviously, it takes a very long time to obtain accurate recognition results. The detection requirements of the present invention are different from those of the prior art. It only needs to detect whether it exists. Therefore, the present invention improves the prior art and calculates the size of the recognition area based on the recognition results of the h-2 frame and the h-1 frame, so that the size of the recognition area can be closer to the actual size of the employee in the h frame, thereby avoiding the recognition area being too small to contain enough detail information, and avoiding the recognition area being too large, increasing the computational pressure for recognizing a single recognition area, effectively improving the accuracy of recognition and reducing the computing power requirements for recognition.

[0119] Preferably, based on (x pre ,y pre ) Calculate the size of the recognition area, including:

[0120] Use Lstd h-1 and Lstd h-2 represents the difference between the maximum and minimum values ​​of the ordinates of the pixel points belonging to the employee in the h-1th frame and the h-2th frame;

[0121] Use Wstd h-1 and Wstd h-2 represents the difference between the maximum and minimum values ​​of the horizontal coordinates of the pixel points belonging to the employee in the h-1th frame and the h-2th frame;

[0122] Then in the hth frame, the calculation formula for the length of the recognition area is:

[0123]

[0124] The calculation formula for the width of the recognition area is:

[0125]

[0126] L h and W h Respectively represent the length and width of the recognition area in the hth frame; AL h and A.W.h Respectively represent the length and width of the hth frame,

[0127] In the hth frame, the size of the recognition area can be changed as (x pre ,y ere ), (x pre ,y pre ) is the present invention according to (x h-1 ,y h-1 ) and (x h-2 ,y h-2 ) when the employee's movement is continuous, within a very small time interval, the movement directions between three adjacent frames are generally the same. In this way, the position relationship between the first two frames can be used to predict the position of the third frame, so that the size of the recognition area can be calculated based on the predicted position, so that the size of the recognition area can change with the positional relationship between the employee and the video capture device, thereby avoiding employee detection based on recognition areas of various sizes and improving the success rate of obtaining detection results.

[0128] When calculating the length and width, the farther the employee's vertical coordinate is from the bottom of the h-th frame, and the farther the horizontal coordinate is from the center of the bottom of the h-th frame, the smaller the detected length and width are, indicating that the distance between the employee and the video capture device is farther, thereby reducing the area of ​​the recognition area to improve the recognition efficiency of a single recognition area.

[0129] Preferably, calculating the detection value of the pixel point in the h-th frame based on the h-1-th frame and the h-2-th frame includes:

[0130] Use b to represent the pixel in the hth frame, and use check b Indicates the detection value of b, then check b The calculation formula is as follows:

[0131]

[0132] dstna b Represents the coordinates of b and the coordinates (x pre ,y pre ), dstna ma Represents the pixel point and coordinate (x pre ,y pre ) is the maximum distance between b Represents b and ray FL h-1 The distance between ma Represents the pixel point in the hth frame and the ray FL h-1The maximum value of the distance between, η1 represents the first detection coefficient, η2 represents the second detection coefficient; FL h-1 For (x h-2 ,y h-2 ) is the endpoint and passes through (x h-1 ,y h-1 ) rays.

[0133] The detection value of the present invention is obtained by weighting two different types of distances. If b and the coordinate (x pre ,y pre ) is closer, and the h-1 The closer the distance between b and the hth frame, the greater the detection value, indicating that the closer the distance between b and the employee in the hth frame is, the greater the possibility is, and the earlier the recognition area centered on b is recognized. In addition, the present invention is not only based on the coordinates of b and the coordinates (x pre ,y pre ) to determine the detection value of b, because there is also a point in the opposite direction of the employee's movement that is different from the coordinate (x pre ,y pre ) are very small, so if we only use the coordinates of b and the coordinates (x pre ,y pre ) to determine the detection value of b, it is easy to make the detection value of the pixel points at these positions too large, so that the recognition area of ​​the pixel points in the area that does not contain employees is placed in the earlier recognition position, which increases the total time for detecting the frame to obtain the recognition result when there is an employee in the frame. Therefore, the present invention also makes a ray FL h-1 ,When calculating the detection value, the size between the pixel and the ray is also considered, so as to further ensure the validity of the detection value.

[0134] Preferably, the first detection coefficient is 0.66 and the second detection coefficient is 0.33.

[0135] Preferably, the total time that the employee is in the clinical trial site in the statistical monitoring video includes:

[0136] Let T1 represent the total duration of the surveillance video, N represent the total number of frames contained in the surveillance video, and M represent the total number of frames of employees present. The calculation formula for the total duration of employees in the clinical trial site is as follows:

[0137]

[0138] T2 represents the total time employees spend at the clinical trial site.

[0139] It is obvious to those skilled in the art that the present invention is not limited to the details of the above-mentioned exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention; therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, and therefore it is intended to include all changes within the meaning and scope of the equivalent elements of the claims within the present invention, and any figure mark in the claims should not be regarded as limiting the claims involved.

[0140] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. Clinical trial project work time analysis management system, It is characterized in that It includes a working time analysis and management device, which includes a video analysis module, and the video analysis module includes an acquisition unit, an analysis unit and a statistical unit; The acquisition unit is used to save all frames contained in the surveillance video into a set FraSet in the order of shooting time from morning to night; The analysis unit is used to determine whether each frame in FraSet contains employees who are present, including: The process of identifying the first and second frames is as follows: The second detection algorithm is used to identify the hth frame to determine whether there is an employee on duty in the first frame; The process of identifying the hth frame is as follows: h is a positive integer greater than or equal to 3, and it is determined whether there are employees present in both the h-1th frame and the h-2th frame. If so, the first detection algorithm is used to identify the hth frame to determine whether there are employees present in the hth frame; if not, the second detection algorithm is used to identify the hth frame to determine whether there are employees present in the hth frame; The first detection algorithm includes: The first step is to calculate the size of the recognition area based on the h-1th frame and the h-2th frame; The second step is to calculate the detection value of the pixel point in the hth frame based on the h-1th frame and the h-2th frame; The third step is to save all the pixels in the hth frame into the set PixSet in descending order of detection values. h ; Step 4: Initialize the value of k to 1; Step 5: PixSet h The kth pixel point in is taken as the center of the recognition area, and the recognition area is identified to determine whether there is an employee on duty in the recognition area. If yes, the detection is stopped, otherwise, the sixth step is entered; Step 6: Update the value of k, add 1 to the value of k, and go to step 7; Step 7: Determine whether the value of k is greater than PixSet h The total number of pixels in , if yes, it means that there is no employee present in the hth frame, if no, go to step 5; The statistical unit is used to count the total time that employees are in the clinical trial site in the surveillance video.

2. The clinical trial project work time analysis and management system according to claim 1, It is characterized in that It also includes a working time acquisition device, which is used to acquire the working hours of employees of the clinical trial project on the attendance day.

3. The clinical trial project work time analysis and management system according to claim 2, It is characterized in that The working time analysis and management device also includes a judgment module and a working time management module; The judgment module is used to compare the total time with the working hours to determine whether the working hours are abnormal; The working hours management module is used by working hours management personnel to correct abnormal working hours and obtain corrected working hours.

4. The clinical trial project work time analysis and management system according to claim 1, It is characterized in that Also included is a video shooting device, which is used to obtain monitoring videos of the clinical trial site of employees of the clinical trial project on attendance days.

5. The clinical trial project work time analysis and management system according to claim 1, It is characterized in that The second detection algorithm includes: Taking each pixel point in the hth frame as the center of a recognition area of ​​a preset size, a plurality of recognition areas are obtained; Identify each identification area separately to determine whether there are employees present in all the identification areas.

6. The clinical trial project work time analysis management system according to claim 1 or 5, It is characterized in that When identifying the recognition area, the R-CNN algorithm or the YOLO algorithm is used to identify the recognition area.

7. The clinical trial project work time analysis and management system according to claim 1, It is characterized in that The size of the recognition area is calculated based on the h-1th frame and the h-2th frame, including: Use (x h-1 ,y h-1 ) represents the coordinates of the center of the recognition area with the most pixels in the area containing employees on duty in the h-1th frame; Use (x h-2 ,y h-2 ) represents the coordinates of the center of the recognition area containing the most pixels belonging to employees on duty in the h-2th frame; In the hth frame, make a h-1 ,y h-1 ) and (x h-2 ,y h-2 ) of the straight line L h-1 ; Calculate (x h-1 ,y h-1 ) and (x h-2 ,y h-2 ) Distance D h-1,h-2 ; Get except (x h-2 ,y h-2 ), and (x h-1 ,y h-1 ) is equal to D h-1,h-2 And through L h-1 The coordinates of the point (x pre ,y pre ); Based on (x pre ,y pre ) Calculate the size of the recognition area.

8. The clinical trial project work time analysis and management system according to claim 7, It is characterized in that Based on (x pre ,y pre ) Calculate the size of the recognition area, including: Use Lstd h-1 and Lstd h-2 represents the difference between the maximum and minimum values ​​of the ordinates of the pixel points belonging to the employee in the h-1th frame and the h-2th frame; Use Wstd h-1 and Wstd h-2 represents the difference between the maximum and minimum values ​​of the horizontal coordinates of the pixel points belonging to the employee in the h-1th frame and the h-2th frame; Then in the hth frame, the calculation formula for the length of the recognition area is: The calculation formula for the width of the recognition area is: L h and W h Respectively represent the length and width of the recognition area in the hth frame; AL h and A.W. h Respectively represent the length and width of the hth frame, 9. The clinical trial project work time analysis and management system according to claim 8, It is characterized in that Calculating the detection value of the pixel point in the hth frame based on the h-1th frame and the h-2th frame, including: Use b to represent the pixel in the hth frame, and use check b Indicates the detection value of b, then check b The calculation formula is as follows: dstna b Represents the coordinates of b and the coordinates (x pre ,y pre ), dstna ma Represents the pixel point and coordinate (x pre ,y pre ) is the maximum distance between b Represents b and ray FL h-1 The distance between ma Represents the pixel point in the hth frame and the ray FL h-1 The maximum value of the distance between, η1 represents the first detection coefficient, η2 represents the second detection coefficient; FL h-1 For (x h-2 ,y h-2 ) is the endpoint and passes through (x h-1 ,y h-1 ) rays.

10. The clinical trial project work time analysis and management system according to claim 8, It is characterized in that Count the total time employees spend in clinical trial sites as recorded in surveillance videos, including: Let T1 represent the total duration of the surveillance video, N represent the total number of frames contained in the surveillance video, and M represent the total number of frames of employees present. The calculation formula for the total duration of employees in the clinical trial site is as follows: T2 represents the total time employees spend at the clinical trial site.

Images