Drowsiness detection device and program for detecting drowsiness

Abstract

To provide a drowsiness detector and a program for the drowsiness detection capable of improving accuracy of drowsiness detection.SOLUTION: A PERCLOS calculation unit 33 calculates a PERCLOS, which is a time ratio that a driver is closing his or her eyes for a fixed time Ta, on the basis of an image captured by a camera 2. A drowsiness detection unit 34 detects drowsiness of the driver when the PERCLOS is a threshold or more. A threshold setting unit 35 sets the threshold lower when the ratio of an initial deep blink calculated first from the start of driving is less than a% than when the ratio of the initial deep blink is a% or more.SELECTED DRAWING: Figure 6

Description

[Technical field]

[0001] The present invention relates to a drowsiness detection device and a program for drowsiness detection. [Background technology]

[0002] PERCLOS (PERcent of eye lid CLOSure) is one of the indices for measuring the degree of fatigue of a driver of a vehicle such as an automobile (Patent Document 1). PERCLOS is the percentage of time that the driver has his / her eyes closed per minute. Drowsiness can be detected when this PERCLOS is equal to or greater than a threshold value. However, there is an issue that PERCLOS when a driver feels drowsy varies from person to person, and if the threshold value is set uniformly, drowsiness cannot be detected with high accuracy.

[0003] Therefore, it is conceivable to apply the technology described in Patent Document 2 so that the driver can change the threshold value at will. However, when the driver changes the threshold value by himself, he needs to check the past PERCLOS history and the threshold value, which is troublesome and may result in the threshold value not being changed or not being set to an appropriate value. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2022-42248 [Patent Document 2] Japanese Patent Application Publication No. 10-40481 Summary of the Invention [Problem to be solved by the invention]

[0005] The present invention has been made in consideration of the above-mentioned circumstances, and has an object to provide a drowsiness detection device and a drowsiness detection program that improve the accuracy of drowsiness detection. [Means for solving the problem]

[0006] In order to achieve the above-mentioned object, a drowsiness detection device according to the present invention has the following features. An imaging unit that captures an image of a face of a person to be detected; a first calculation unit that calculates a drowsiness index value that is a time during which the subject has his / her eyes closed during a first predetermined time or a value corresponding to the time based on the image captured by the image capture unit; a drowsiness detection unit that detects drowsiness of the detection subject when the drowsiness index value is equal to or greater than a threshold value. In the drowsiness detection device, A second calculation unit that calculates a total number of blinks at which the eye opening degree becomes equal to or smaller than a first predetermined value during a second predetermined time based on the image captured by the image capture unit; a third calculation unit that calculates the number of deep blinks at which the eye opening degree is equal to or less than a second predetermined value lower than the first predetermined value during the second predetermined time based on the image captured by the imaging unit; A fourth calculation unit that calculates a ratio of deep blinks, which is a ratio of the number of deep blinks to the total number of blinks; A threshold setting unit that sets the threshold based on the initial deep blink rate calculated for the first time after a predetermined trigger. It is a drowsiness detection device.

[0007] In order to achieve the above-mentioned object, a drowsiness detection program according to the present invention has the following features. On the computer, a first calculation unit that calculates a drowsiness index value that is a time during which the detection subject has his / her eyes closed during a first predetermined time or a value corresponding to the time, based on an image of the detection subject's face captured by an image capture unit; a drowsiness detection unit that detects drowsiness of the detection subject when the drowsiness index value is equal to or greater than a threshold value; A second calculation unit that calculates a total number of blinks at which the eye opening degree becomes equal to or smaller than a first predetermined value during a second predetermined time based on the image captured by the image capture unit; a third calculation unit that calculates the number of deep blinks at which the eye opening degree is equal to or less than a second predetermined value lower than the first predetermined value during the second predetermined time based on the image captured by the imaging unit; A fourth calculation unit that calculates a ratio of deep blinks, which is a ratio of the number of deep blinks to the total number of blinks; and functioning as a threshold setting unit that sets the threshold based on a ratio of the initial deep blink calculated for the first time after a predetermined trigger. It is a program for detecting drowsiness. Effect of the Invention

[0008] According to the present invention, it is possible to provide a drowsiness detection device and a drowsiness detection program that improve the accuracy of drowsiness detection.

[0009] The present invention has been briefly described above. Furthermore, the details of the present invention will be further clarified by reading the following description of the embodiment of the present invention (hereinafter, referred to as "embodiment") with reference to the accompanying drawings. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a block diagram showing a drowsiness detection device according to a first embodiment of the present invention. [Diagram 2] FIG. 2 is a time chart for explaining the function of the eye closure detection unit shown in FIG. [Diagram 3] FIG. 3 is a time chart for explaining the function of the PERCLOS calculation unit shown in FIG. [Figure 4] FIG. 4 is a time chart showing the opening and closing of the eyes of a person who is blinking deeply. [Diagram 5] FIG. 5 is a time chart showing the opening and closing of the eyes of a person who blinks shallowly. [Figure 6] FIG. 6 is a flowchart showing the operation of the drowsiness detection device shown in FIG. [Figure 7] FIG. 7 is a time chart showing PERCLOS for a person with a normal blink rate and a person with a low blink rate. [Figure 8] FIG. 8 is a block diagram showing a drowsiness detection device according to the second embodiment of the present invention. [Figure 9]FIG. 9 is a flowchart showing the operation of the drowsiness detection device shown in FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Specific embodiments of the present invention will be described below with reference to the accompanying drawings.

[0012] The drowsiness detection device 1 of the present embodiment shown in Fig. 1 is a device that is mounted in a vehicle and detects drowsiness of a driver who is a detection subject seated in the driver's seat. As shown in the figure, the drowsiness detection device 1 includes a camera 2 as an imaging unit, a control unit 3, and an alarm output unit 4.

[0013] The camera 2 is installed in the meter unit in front of the driver's seat or on the column cover so that it can capture an image of the face of the driver sitting in the driver's seat. The camera 2 supplies the captured image of the driver's face to the control unit 3, which will be described later.

[0014] The control unit 3 is composed of a well-known microcomputer composed of a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), etc., and functions as a computer that operates according to a program for drowsiness detection. The control unit 3 has a face detection unit 31, a closed-eye detection unit 32, a PERCLOS calculation unit 33 as a first calculation unit, a drowsiness detection unit 34, a threshold setting unit 35, a total blink calculation unit 36 ​​as a second calculation unit, a deep blink calculation unit 37 as a third calculation unit, and a deep blink proportion calculation unit 38 as a fourth calculation unit.

[0015] The face detection unit 31 extracts a face image from an image captured by the camera 2. The closed eye detection unit 32 obtains the degree of eye opening and closing from the face image detected by the face detection unit 31, and detects closed eyes (blinks) from the degree of eye opening and closing. In this embodiment, as shown in FIG. 2, the closed eye detection unit 32 detects blinks ranging from shallow blinks with an opening and closing degree of 50% (first predetermined value) or less, when 100% opening and closing is fully open and 0% opening and closing is fully closed, and supplies the results to the total blink calculation unit 36. The closed eye detection unit 32 also detects deep blinks with an opening and closing degree of 20% (second predetermined value) or less, and supplies the results to the PERCLOS calculation unit 33 and the deep blink calculation unit 37.

[0016] As shown in FIG. 3, the PERCLOS calculation unit 33 calculates PERCLOS shown in the following formula (1) as a drowsiness index value based on the deep blink detection result from the eye closure detection unit 32. PERCLOS=(T1+T2+T3) / Ta×100% …(1)

[0017] PERCLOS is the ratio of the time (T1+T2+T3) that the driver has his / her eyes closed to the given time Ta (=first predetermined time). The PERCLOS calculation unit 33 calculates PERCLOS for each given time Ta.

[0018] The drowsiness detection unit 34 detects the driver's drowsiness when PERCLOS is equal to or greater than a threshold, and causes the warning output unit 4 to generate a warning. The threshold setting unit 35 will be described later. The total blink calculation unit 36 ​​calculates the number of blinks with an opening degree of 50% or less in a certain time Ta (a second predetermined time) as the total blink count n1, and supplies the calculation result to the deep blink proportion calculation unit 38. The deep blink calculation unit 37 calculates the number of deep blinks with an opening degree of 20% or less in the certain time Ta as the deep blink count n2, and supplies the calculation result to the deep blink proportion calculation unit 38.

[0019] The deep blink proportion calculation unit 38 calculates the proportion of the number of deep blinks n2 to the total number of blinks n1 as the proportion of deep blinks, as shown in the following formula (2). Proportion of deep blinks = n2 / n1 …(2)

[0020] For example, a person with deep blinks as shown in Figure 4 has 100% deep blinks because all blinks are deep blinks with an opening / closing ratio of 20% or less. In contrast, a person with shallow blinks as shown in Figure 5 has two deep blinks with an opening / closing ratio of 20% or less, as shown in the dotted circle, and a total of 14 blinks with an opening / closing ratio of 50% or less, as shown in the dotted and solid circles. Therefore, the percentage of deep blinks is low at 14.3%, as shown in the following formula (3). Percentage of deep blinks = 2 / 14 = 14.3% … (3)

[0021] The threshold setting unit 35 sets the threshold based on the ratio of deep blinks calculated in the first certain time Ta after the start of driving is detected (after a predetermined trigger).

[0022] The warning output unit 4 is composed of a display device such as a meter display, a navigation monitor, or a head-up display, and notifies the driver by sound, voice, text, icon, or the like that drowsiness has been detected or that a break is recommended.

[0023] Next, the operation of the drowsiness detection device 1 described above will be described with reference to the flowchart in Fig. 6. First, when the control unit 3 detects the start of driving, it resets the threshold value used for the above-mentioned drowsiness detection (i.e., sets it to the initial state). The control unit 3 detects the start of driving, for example, when the ignition switch is turned on.

[0024] The control unit 3 repeatedly performs the drowsiness detection process shown in Fig. 6 every time one frame of image captured by the camera 2 is supplied. In the drowsiness detection process, the control unit 3 extracts a face image from the image captured by the camera 2 (S1). Next, the control unit 3 obtains the eye opening degree from the face image detected by the face detection unit 31, and detects blinks with an opening degree of 50% or less and blinks with an opening degree of 20% or less based on the opening degree (S2). Next, when a blink is detected in S2, the control unit 3 counts up the time the driver's eyes are closed, and when a certain time Ta has passed, calculates PERCOLS based on the counted up time (S3).

[0025] If the fixed time Ta has not elapsed and PERCLOS has not been calculated (N in S4), the control unit 3 ends the process. If the fixed time Ta has elapsed and PERCLOS has been calculated (Y in S4), the control unit 3 determines whether or not a threshold has been set (S5). If a threshold has been set (Y in S5), the control unit 3 immediately proceeds to S9.

[0026] If the threshold is not set (N in S5), the control unit 3 determines whether the percentage of early deep blinks calculated in the first certain time Ta after starting driving is equal to or greater than a% (third predetermined value) (S6). If the percentage of early deep blinks is equal to or greater than a% (Y in S6), the control unit 3 sets the threshold to C1% (S8) and proceeds to S9. If the percentage of early deep blinks is less than a% (N in S6), the control unit 3 sets the threshold to C2%, which is lower than C1%, and proceeds to S9 (S7).

[0027] In S9, if PERCLOS calculated in S3 is equal to or greater than the threshold (Y in S9), the control unit 3 determines that the subject is drowsy (S11), controls the alarm output unit 4 to generate an alarm, and then ends the process. If PERCLOS calculated in S3 is less than the threshold (N in S9), the control unit 3 determines that the subject is not drowsy (S10), and ends the process.

[0028] As shown in Fig. 4, people who blink deeply have high PERCLOS, and as shown in Fig. 5, people who blink shallowly do not detect closed eyes and have low PERCLOS. As shown in Fig. 7, people who blink infrequently or shallowly do not have as high a PERCLOS as people who blink normally, even if they become sleepy and blink more frequently. If the threshold is set uniformly, for example at 12%, as in the conventional technology, it is not possible to detect drowsiness in people who blink infrequently or shallowly.

[0029] According to the above-described embodiment, when the percentage of early deep blinks is less than a%, the control unit 3 sets the threshold lower than when the percentage of early deep blinks is a% or more (C1>C2). That is, therefore, for a person who blinks shallowly, the percentage of early deep blinks is less than a%, and the threshold can be set lower than the threshold for a person who blinks normally. This allows a threshold appropriate for the driver to be set, and improves the accuracy of drowsiness detection.

[0030] According to the above-described embodiment, the detection of the start of driving is set as the predetermined trigger. This makes it possible to determine the threshold value based on the rate of early deep blinks close to those at the time of awakening, thereby further improving the accuracy of drowsiness detection.

[0031] Second embodiment Next, a drowsiness detection device 1 of a second embodiment will be described with reference to Fig. 8. In Fig. 8, parts equivalent to those of the drowsiness detection device 1 shown in Fig. 1 already described in the first embodiment are given the same reference numerals and detailed description will be omitted. A difference between the drowsiness detection device 1B of the second embodiment and the drowsiness detection device 1 of the first embodiment is that a threshold setting unit 35 sets a threshold based on the proportion of deep blinks and the number of deep blinks calculated in the first certain time Ta after driving starts.

[0032] Next, the operation of the drowsiness detection device 1 of the second embodiment will be described with reference to the flowchart of Fig. 9. In Fig. 9, steps that perform the same operations as those in the flowchart shown in Fig. 6 already described in the above-mentioned first embodiment are omitted, or the same reference numerals are used and detailed descriptions thereof are omitted.

[0033] In the drowsiness detection process, the control unit 3 executes the operations of S1 to S6 in the same manner as in the first embodiment. In S6, the control unit 3 determines whether the percentage of early deep blinks is a% (predetermined value) or more (S6). If the percentage of early deep blinks is a% or more (Y in S6), the control unit 3 sets the threshold to C1% (S8) and then proceeds to S9. If the percentage of early deep blinks is less than a% (N in S6), the control unit 3 determines whether the number of early blinks calculated in the first certain time Ta after starting driving is b times (predetermined number) or more (S12).

[0034] If the blink count is b or more (Y in S12), the control unit 3 sets the threshold to C2%, which is lower than C1%, (S7) and then proceeds to S9. If the blink count is less than b (N in S12), the control unit 3 sets the threshold to C3%, which is lower than C2%, (S13) and then proceeds to S9. Since the process from S9 onwards is the same as in the first embodiment, the description will be omitted.

[0035] For people whose blink counts are measured to a certain extent, PERCLOS increases slightly when they become more sleepy, but for people whose blink counts are barely measured, PERCLOS does not increase much even when they become more sleepy.

[0036] According to the above-described embodiment, when the ratio of early deep blinks is less than a% and the number of early blinks is less than b, the control unit 3 sets the threshold lower than when the ratio of early deep blinks is less than a% and the number of early blinks is b or more (C2>C3). As a result, a person who blinks less often will have a blink count less than b, and the threshold can be set lower than a person who blinks more often. This can further improve the accuracy of drowsiness detection. C2 and C3 may be calculated using the following formula with the ratio of deep blinks (α) and the total number of blinks (β) as parameters. Threshold = (n / β) × α … (2)

[0037] Next, in order to confirm the effects of the first and second embodiments, the inventors measured the proportion of early deep blinks two times for each of subjects A to D (detection subjects), and had them self-report their drowsiness level when they became more drowsy.The threshold values ​​and detection results of the drowsiness detection devices 1 and 1B of the conventional, first, and second embodiments are summarized in Table 1 below.

[0038] [Table 1]

[0039] In the example shown in Table 1, a=70%, b=15 times, C1=12%, C2=5%, and C3=3%. As is clear from Table 1, in the first and second blinks of subject C, in which the rate of initial deep blinks is less than 70%, and in the first and second blinks of subject D, drowsiness could not be detected even if the subject felt drowsy in the conventional technology in which the threshold value is uniformly set to 12%. In contrast, it was found that the drowsiness detection device 1 of the first embodiment can detect drowsiness in the second blink of subject C, and the accuracy of drowsiness detection is improved. It was found that the drowsiness detection device 1B of the second embodiment can detect drowsiness in the first and second blinks of subject C, and the second blink of subject D, and the accuracy of drowsiness detection is further improved. When using formula (2) with n = 9, the detection results for the first and second times for subject C and the first and second times for subject D were as follows: first time threshold for subject C: (9 / 33) × 12.1% = 3.3, second time for subject C: 10.6, first time for subject D: 2.1. Since the second time for subject D cannot be calculated, if the lower limit of the threshold (for example, 3) is set, the detection rate becomes 90, indicating that accuracy is improved.

[0040] The present invention is not limited to the above-described embodiment, and can be appropriately modified, improved, etc. In addition, the material, shape, size, number, arrangement location, etc. of each component in the above-described embodiment are arbitrary as long as the present invention can be achieved, and are not limited.

[0041] According to the above-described embodiment, the drowsiness index value is calculated as the percentage of time that the driver has his / her eyes closed during the fixed time Ta, but the present invention is not limited to this. The drowsiness index value may be calculated as the integral of the time that the driver has his / her eyes closed during the fixed time Ta.

[0042] According to the above-mentioned embodiment, the control unit 3 detects the start of driving when the ignition switch is turned on and resets the drowsiness detection threshold, but this is not limited to the above. For example, since the driver may be changed with the engine running, the drowsiness detection threshold may be reset when the driver performs a reset operation (predetermined trigger), or the threshold may be remeasured at regular intervals after the start of driving and updated if there is a change.

[0043] According to the embodiment described above, the fixed time Ta for calculating PERCLOS is the same as the fixed time Ta for calculating the percentage of deep blinks and the number of blinks, but this is not limited to this and may be different times.

[0044] Here, features of the embodiments of the drowsiness detection device and the drowsiness detection program according to the present invention described above will be briefly summarized and listed in the following [1] to [6]. [1] An imaging unit (2) that captures an image of the face of a person to be detected; a first calculation unit (33) that calculates a drowsiness index value that is a time during which the subject has his / her eyes closed during a first predetermined time or a value corresponding to the time, based on the image captured by the imaging unit (2); a drowsiness detection unit (34) that detects drowsiness of the detection subject when the drowsiness index value is equal to or greater than a threshold value; In a drowsiness detection device (1), a second calculation unit (36) that calculates a total number of blinks at which the eye opening / closing degree is equal to or smaller than a first predetermined value during a second predetermined time based on the image captured by the image capture unit (2); a third calculation unit (37) that calculates the number of deep blinks at which the eye opening degree becomes equal to or less than a second predetermined value lower than the first predetermined value during the second predetermined time based on the image captured by the imaging unit (2); A fourth calculation unit (38) that calculates a ratio of deep blinks, which is a ratio of the number of deep blinks to the total number of blinks; A threshold setting unit (35) that sets the threshold based on the proportion of the initial deep blink first calculated after a predetermined trigger. Drowsiness detection device (1).

[0045] According to the above configuration [1], a threshold suitable for the driver can be set, and the accuracy of drowsiness detection can be improved.

[0046] [2] In the drowsiness detection device (1) according to [1], The threshold setting unit (35) sets the threshold lower when the ratio of the deep blinks at the initial stage is less than a third predetermined value than when the ratio of the deep blinks at the initial stage is equal to or greater than the third predetermined value. Drowsiness detection device (1).

[0047] According to the configuration of [2] above, for a person who blinks shallowly, the proportion of initial deep blinks is less than the third predetermined value, and the threshold can be set lower than the threshold for a person who blinks normally. This makes it possible to set a threshold that is more suitable for the driver, thereby improving the accuracy of drowsiness detection.

[0048] [3] In the drowsiness detection device (1) according to [1], The threshold setting unit (35) sets the threshold based on the ratio of the initial deep blink and the number of the initial deep blinks calculated for the first time after the predetermined trigger. Drowsiness detection device (1).

[0049] According to the above configuration [3], a threshold value that is more suitable for the driver can be set, and the accuracy of drowsiness detection can be improved.

[0050] [4] In the drowsiness detection device (1) according to [3], The threshold setting unit (35) When the proportion of the initial deep blink is less than a third predetermined value, the threshold is set lower than when the proportion of the initial deep blink is equal to or greater than the third predetermined value; When the proportion of deep blinks in the initial stage is less than the third predetermined value and the number of deep blinks in the initial stage is less than a predetermined number, the threshold value is set lower than when the proportion of deep blinks in the initial stage is less than the third predetermined value and the number of deep blinks in the initial stage is equal to or greater than the predetermined number. Drowsiness detection device (1).

[0051] According to the configuration of [4] above, the number of blinks of a person who blinks infrequently is less than the predetermined number, and the threshold value can be set lower than that of a person who blinks frequently. This makes it possible to further improve the accuracy of drowsiness detection.

[0052] [5] In the drowsiness detection device (1) according to any one of [1] to [4], Detection of the start of operation is set as the predetermined trigger. Drowsiness detection device (1).

[0053] According to the above configuration [5], the threshold value can be determined based on the initial drowsiness index value close to the time of awakening, thereby further improving the accuracy of drowsiness detection.

[0054] [6] On the computer, a first calculation unit (33) that calculates a drowsiness index value, which is a time during which the detection subject has his / her eyes closed during a first predetermined time or a value corresponding to the time, based on an image of the detection subject's face captured by the capture unit (2); a drowsiness detection unit (34) that detects drowsiness of the detection subject when the drowsiness index value is equal to or greater than a threshold value; a second calculation unit (36) that calculates a total number of blinks at which the eye opening / closing degree is equal to or smaller than a first predetermined value during a second predetermined time based on the image captured by the image capture unit (2); a third calculation unit (37) that calculates the number of deep blinks at which the eye opening degree becomes equal to or less than a second predetermined value lower than the first predetermined value during the second predetermined time based on the image captured by the imaging unit (2); A fourth calculation unit (38) that calculates a ratio of deep blinks, which is a ratio of the number of deep blinks to the total number of blinks; a threshold setting unit (35) that sets the threshold based on the proportion of the initial deep blink that is calculated for the first time after a predetermined trigger; A program for detecting drowsiness.

[0055] According to the configuration [6] above, a threshold suitable for the driver can be set, thereby improving the accuracy of drowsiness detection. [Explanation of symbols]

[0056] 1. Drowsiness detection device 2 Camera (photography section) 33 PERCLOS calculation unit (first calculation unit) 34 Drowsiness detection unit 35 Threshold setting unit 36 Total blink calculation unit (second calculation unit) 37 Deep blink calculation unit (third calculation unit) 38 Deep blink rate calculation unit (fourth calculation unit)

Claims

1. An imaging unit that captures an image of a face of a person to be detected; a first calculation unit that calculates a drowsiness index value that is a time during which the subject has his / her eyes closed during a first predetermined time or a value corresponding to the time, based on the image captured by the image capture unit; a drowsiness detection unit that detects drowsiness of the detection subject when the drowsiness index value is equal to or greater than a threshold value. In the drowsiness detection device, A second calculation unit that calculates a total number of blinks at which the eye opening degree becomes equal to or smaller than a first predetermined value during a second predetermined time based on the image captured by the image capture unit; a third calculation unit that calculates the number of deep blinks at which the eye opening degree is equal to or less than a second predetermined value lower than the first predetermined value during the second predetermined time based on the image captured by the imaging unit; A fourth calculation unit that calculates a ratio of deep blinks, which is a ratio of the number of deep blinks to the total number of blinks; A threshold setting unit that sets the threshold based on the initial deep blink rate calculated for the first time after a predetermined trigger. Drowsiness detection device.

2. 2. The drowsiness detection device according to claim 1, The threshold setting unit sets the threshold lower when the ratio of the deep blink at the initial stage is less than a third predetermined value than when the ratio of the deep blink at the initial stage is equal to or greater than the third predetermined value. Drowsiness detection device.

3. 2. The drowsiness detection device according to claim 1, The threshold setting unit sets the threshold based on the ratio of the initial deep blink and the number of the initial deep blinks calculated for the first time after the predetermined trigger. Drowsiness detection device.

4. The drowsiness detection device according to claim 3, The threshold setting unit is When the proportion of the initial deep blink is less than a third predetermined value, the threshold is set lower than when the proportion of the initial deep blink is equal to or greater than the third predetermined value; When the ratio of the deep blinks at the initial stage is less than the third predetermined value and the number of the deep blinks at the initial stage is less than a predetermined number, the threshold value is set lower than when the ratio of the deep blinks at the initial stage is less than the third predetermined value and the number of the deep blinks at the initial stage is equal to or greater than the predetermined number. Drowsiness detection device.

5. The drowsiness detection device according to any one of claims 1 to 4, Detection of the start of operation is set as the predetermined trigger. Drowsiness detection device.

6. On the computer, a first calculation unit that calculates a drowsiness index value that is a time during which the detection subject has his / her eyes closed during a first predetermined time or a value corresponding to the time, based on an image of the detection subject's face captured by the image capture unit; a drowsiness detection unit that detects drowsiness of the detection subject when the drowsiness index value is equal to or greater than a threshold value; A second calculation unit that calculates a total number of blinks at which the eye opening degree becomes equal to or smaller than a first predetermined value during a second predetermined time based on the image captured by the image capture unit; a third calculation unit that calculates the number of deep blinks at which the eye opening degree is equal to or less than a second predetermined value lower than the first predetermined value during the second predetermined time based on the image captured by the imaging unit; A fourth calculation unit that calculates a ratio of deep blinks, which is a ratio of the number of deep blinks to the total number of blinks; and functioning as a threshold setting unit that sets the threshold based on a ratio of the initial deep blink calculated for the first time after a predetermined trigger. A program for detecting drowsiness.

Images