Vehicle driver real-time state identification and evaluation method

Abstract

The invention relates to a vehicle driver real-time state identification and evaluation method, and the method comprises the steps: obtaining a user image set of a target user in a preset time period, and enabling the user image set to comprise a plurality of user images; performing state recognition on a local part of the face of the target user in the user image to obtain a user local state corresponding to the user image; according to an image acquisition sequence, state change identification is carried out on user local states corresponding to the user images in the user image set in sequence, a continuous state change result is obtained, and the image acquisition sequence is a sequence corresponding to user image acquisition; and determining the user state of the target user according to the continuous state change result, and finally obtaining an identification evaluation result. According to the vehicle driver real-time state identification and evaluation method, the accuracy of state identification can be improved, and misinformation of terminal equipment can be filtered out.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field related to road traffic safety, and particularly relates to a real-time state recognition and evaluation method for a vehicle driver. BACKGROUND

[0002] In recent years, with the rapid increase in the number of vehicles on the road, traffic accidents occur frequently. A large part of traffic accidents is caused by driver fatigue driving. The existing fatigue driving detection method includes analyzing the image of the driver. However, when analyzing the image of the driver, the habitual behavior of the driver will interfere, resulting in inaccurate analysis results. SUMMARY

[0003] The purpose of the present application is to provide a real-time state recognition and evaluation method for a vehicle driver, which improves the accuracy of state recognition and filters out false positives of terminal equipment.

[0004] To achieve the above purpose, the present application provides the following technical scheme: a real-time state recognition and evaluation method for a vehicle driver, specifically comprising the following steps: Step S110: obtaining a user image set of a target user in a preset time period, the user image set comprising a plurality of user images; Step S120: performing state recognition on a local part of the face of the target user in the user image to obtain a user local state corresponding to the user image; Step S130: sequentially performing state change recognition on the user local state corresponding to the user image in the user image set according to an image acquisition sequence to obtain a continuous state change result, the image acquisition sequence being the sequence corresponding to the acquisition of the user image; Step S140: determining the user state of the target user according to the continuous state change result to obtain a final recognition and evaluation result.

[0005] As a further improvement of the present application, the user state at least includes a first state, and the determination of the user state of the target user according to the continuous state change result comprises: when the continuous state change result meets a preset condition, the user state of the target user is determined as the first state.

[0006] As a further improvement of the present application, the state change recognition of step S130 comprises the following specific steps: Step S131: obtaining mapping relationship information, the mapping relationship information comprising a mapping relationship between the user local state and a state identifier; Step S132: determining the state identifier associated with the user local state corresponding to the user image according to the mapping relationship information to obtain identifier information; Step S133: obtaining an identifier information set according to the user image set and the identifier information; Step S134: constructing a dynamic change graph corresponding to the identification information in the identification information set according to the preset rule and the image acquisition sequence, to obtain the continuous state change result.

[0007] As a further improvement of the present application, the step S134 of constructing a dynamic change graph corresponding to the identification information in the identification information set includes the following specific steps: Step S1341: merging a plurality of continuous and same identification information in the identification information set into one identification information, to obtain new identification information; Step S1342: obtaining a new identification information set according to the new identification information and the identification information set; Step S1343: constructing a dynamic change graph corresponding to the new identification information in the new identification information set according to the preset rule and the image acquisition sequence, to obtain the continuous state change result.

[0008] As a further improvement of the present application, the user local state of the step S120 includes a first local state and a second local state, and the step S140 of determining the user state in which the target user is located includes the following specific steps: Step S1401: obtaining a set of continuous user images having the first local state and the second local state, to obtain a mixed state set; Step S1402: determining the length of the mixed state set in the user image set, to obtain a mixed state length; Step S1403: determining the proportion of the user images corresponding to the first local state and the second local state in the user image set, to obtain a first proportion value and a second proportion value; Step S1404: obtaining a plurality of first local state sets from the user image set, each of which is a set of continuous user images having the first local state; Step S1405: determining the user state in which the target user is located according to the mixed state length, the first proportion value, the second proportion value, and the plurality of first local state sets.

[0009] As a further improvement of the present application, the step S1405 of determining the user state in which the target user is located according to the mixed state length, the first proportion value, the second proportion value, and the plurality of first local state sets includes the following specific steps: Step S14051: performing feature extraction on the user image containing the first local state to obtain a local feature; Step S14052: determining the length pixel number and the width pixel number corresponding to the local feature, the length pixel number being used to represent the number of pixels occupied by the local feature in the length direction, and the width pixel number being used to represent the number of pixels occupied by the local feature in the width direction. Step S14053: determining a local feature ratio value by a ratio between the length pixel number and the width pixel number; Step S14054: determining a user state in which the target user is located according to the local feature ratio value, the mixed state length, the first state proportion ratio value, the second state proportion ratio value and the plurality of first state sets in the user image set.

[0010] Compared with the prior art, the beneficial effects of the present application are: the technical solution sequentially identifies the state changes of the user local state corresponding to the user image of the user image set according to the image acquisition sequence, thereby obtaining a continuous change state, that is, a dynamic change process of the local part in the preset time period, and determining the user state in which the target user is located according to the dynamic change process, which is beneficial to identifying the user state in which the target user is located in the preset time period, avoids determining the user state in which the target user is located only according to a specific user local state, and improves the accuracy of state recognition. DETAILED DESCRIPTION

[0011] The technical solutions in the embodiments of the present application will be described clearly and completely below, and the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0012] The present application provides the following technical solutions.

[0013] The vehicle driver real-time state recognition and evaluation method of the present application can have the following specific process: Obtain a user image set of a target user in a preset time period, and the user image set contains a plurality of user images.

[0014] The target user is a user to be subjected to state recognition. For example, the target user can be a driver to be subjected to state recognition, the target user can also be a law enforcement officer to be subjected to state recognition, the target user can also be a student to be subjected to state recognition, and the like.

[0015] The user image is an image containing the target user.

[0016] The user image set is a set composed of a plurality of user images.

[0017] Identify the state of the local part of the face of the target user in the user image to obtain a user local state corresponding to the user image.

[0018] The face of the target user is the face of the target user.

[0019] The local part of the target user face is a part in the target user face, such as an eye part, a mouth part, and the like.

[0020] The state recognition is used to recognize the state of the local part, for example, the state recognition can be performed by a deep neural network model trained by user images, and the local part of the user images is marked with a corresponding state.

[0021] The user local state is a state corresponding to the local part of the user, for example, when the local part is a mouth part, the user local state can be wide mouth, small mouth, closed mouth. When the local part is an eye part, the user local state can be wide open eyes, small open eyes, closed eyes.

[0022] According to the image acquisition sequence, the state change of the user local state corresponding to the user image in the user image set is recognized in sequence to obtain a continuous state change result. The image acquisition sequence is the sequence of acquiring the user image.

[0023] The image acquisition sequence is the sequence of acquiring the user image.

[0024] The state change recognition is used to recognize the change of the user local state.

[0025] The continuous state change result is used to represent the change process of the user local state in the user image set.

[0026] For example, when the local part of the target user face is a mouth part, the user image set contains 6 user images, and the continuous state change result corresponding to the user local state of the 6 user images is: small mouth, small mouth, wide mouth, wide mouth, small mouth, wide mouth, which does not limit the continuous state change result of the mouth part.

[0027] For example, when the local part of the target user face is an eye part, the user image set contains 6 user images, and the continuous state change result corresponding to the user local state of the 6 user images is: closed eyes, small open eyes, wide open eyes, wide open eyes, small open eyes, closed eyes, which does not limit the continuous state change result of the eye part.

[0028] In order to facilitate state change recognition, the state change of the user local state corresponding to the user image in the user image set is recognized in sequence according to the image acquisition time to obtain a continuous state change result, including: Obtain mapping relationship information, the mapping relationship information includes the mapping relationship between the user local state and the state identifier; According to the mapping relationship information, the state identifier associated with the user local state corresponding to the user image is determined to obtain identification information; According to the user image set and the identification information, an identification information set is obtained; According to the preset rule and the image acquisition sequence, a dynamic change graph corresponding to the identification information in the identification information set is constructed, and a continuous state change result is obtained.

[0029] The state identification is an identification representing a local state of the user. For example, when the local part is a mouth, the state identification corresponding to closing the mouth can be 0, the state identification corresponding to slightly opening the mouth can be 1, and the state identification corresponding to widely opening the mouth can be 2.

[0030] For example, the state identification can also be A, B, and C, respectively corresponding to closing the mouth, slightly opening the mouth, and widely opening the mouth, and can also be respectively corresponding to closing the eyes, slightly opening the eyes, and widely opening the eyes, and the like.

[0031] The identification information refers to an identification corresponding to a local state of the user. For example, if the local state of the user corresponding to the user image is closing the mouth, the identification information is 0; if the local state of the user corresponding to the user image is closing the eyes, the identification information is A, and the like.

[0032] The identification information set is a set composed of identification information corresponding to the user image set.

[0033] For example, the user image set contains 6 frames of user images, and when the continuous state change result of the local state of the user corresponding to the 6 frames of user images is slightly opening the mouth, slightly opening the mouth, widely opening the mouth, widely opening the mouth, slightly opening the mouth, and widely opening the mouth, the identification information set can be (112212), the identification information set can also be (BBCCBC), and the like.

[0034] In some embodiments, in order to consider that there are continuous same states in the identification information set, and in order to reduce the identification difficulty of the identification information set, according to the preset rule and the image acquisition sequence, a dynamic change graph corresponding to the identification information in the identification information set is constructed, and a continuous state change result is obtained, including: The multiple identification information that are continuous and same in the identification information set are merged into one identification information, and a new identification information is obtained; According to the new identification information and the identification information set, a new identification information set is obtained; According to the preset rule and the image acquisition sequence, a dynamic change graph corresponding to the new identification information in the new identification information set is constructed, and a continuous state change result is obtained.

[0035] The new identification information is used to represent multiple continuous and same states. For example, the identification information set corresponding to the user image set is (00000011112222222), the new identification information corresponding to 000000 can be 0, the new identification information corresponding to 1111 can be 1, the new identification information corresponding to 2222222 can be 2, and the like.

[0036] The new identification information set is a set composed of new identification information. For example, the identification information set corresponding to the user image set is (00000011112222222111111), and the new identification information set can be (0121).

[0037] The dynamic change graph is used to reflect the dynamic change process of the user local state. For example, the dynamic change graph can be a waveform graph, and can also be a column graph, etc.

[0038] According to the continuous state change result, the user state in which the target user is located is determined.

[0039] The user state is used to represent the current situation of the user. For example, the user state can be yawning, eye abnormality, fatigue state, non-fatigue state, etc.

[0040] In order to consider that the target user can correspond to multiple types of user states, the user state at least includes a first state, and the user state in which the target user is located is determined according to the continuous state change result, including: When the continuous state change result meets a preset condition, the user state of the target user is determined as the first state.

[0041] The first state is a state corresponding to the target user. For example, the first state can be a fatigue state or a non-fatigue state; it can also be a yawning state or a non-yawning state; it can also be an eye abnormality state or an eye normal state, etc.

[0042] The preset condition is a condition set in advance, which is used to determine the user state corresponding to the continuous state change result.

[0043] In order to consider that the obtained user image set cannot contain the entire change process of the local part, and to accurately identify the user state in which the target user is located, the user local state includes a first local state and a second local state, and the user state in which the target user is located is determined according to the continuous state change result, including: A group of continuous user images with the first local state and the second local state are obtained, and a mixed state set is obtained; The length of the mixed state set in the user image set is determined, and a mixed state length is obtained; The proportion of the user images corresponding to the first local state and the second local state in the user image set is determined, and a first proportion value and a second proportion value are obtained; A plurality of groups of continuous user images with the first local state are obtained in the user image set, and a plurality of first local state sets are obtained. According to the mixed state length, the first proportion value, the second proportion value, and the plurality of first local state sets, a user state in which the target user is located is determined.

[0044] The first local state is used to represent a state of a local position. For example, the first local state can represent that the mouth is wide open, or can represent that the eyes are closed, and the like.

[0045] The second local state is used to represent a state of a local position. For example, the first local state can represent that the eyes are slightly open, or can represent that the eyes are open, and the like.

[0046] The mixed state set is composed of a group of continuous user images containing the first local state and the second local state.

[0047] The group of continuous user images means that each frame of user image in the plurality of frames of user images corresponds to the image acquisition sequence one by one, that is, there is no frame drop between adjacent frames.

[0048] For example, the mixed state set can be composed of the user images of the 1st frame (wide mouth), the 2nd frame (wide mouth), the 3rd frame (small mouth), the 4th frame (small mouth), the 5th frame (small mouth), the 6th frame (wide mouth), the 7th frame (small mouth), and the 8th frame (small mouth).

[0049] For example, the mixed state set can be composed of the user images of the 1st frame (closed eyes), the 2nd frame (closed eyes), the 3rd frame (open eyes), the 4th frame (open eyes), the 5th frame (closed eyes), the 6th frame (closed eyes), the 7th frame (closed eyes), and the 8th frame (open eyes).

[0050] The mixed state length refers to the length of the mixed state set relative to the user image set. For example, the mixed state set has 16 frames of user images, and the user image set has 30 frames of user images, and the mixed state length is 16.

[0051] The first proportion value is the proportion of the user image in which the user local state is the first local state in the user image set. For example, the user image set has 30 frames of user images, and the user image in which the user local state is the first local state has 12 frames, and the first proportion value is 0.4.

[0052] The second proportion value is the proportion of the user image in which the user local state is the second local state in the user image set. For example, the user image set has 30 frames of user images, and the user image in which the user local state is the first local state has 15 frames, and the first proportion value is 0.5.

[0053] The first local state set is composed of user images containing the first local state, and the image acquisition sequence corresponding to the user images in the first local state set is sequentially connected. For example, the user local state in the user image set is: the first frame (closed mouth), the second frame (closed mouth), the third frame (open mouth), the fourth frame (open mouth), the fifth frame (open mouth), the sixth frame (open mouth), the sixth frame (open mouth), the eighth frame (open mouth), the ninth frame (open mouth), and the tenth frame (open mouth). The user image set has two first local state sets, which are (5th frame, 6th frame) and (9th frame, 10th frame).

[0054] For example, the preset condition provides that when the mixed state length is not less than 15, the first proportion value is not less than 0.3, the second proportion value is not more than 0.6, and the number of the plurality of first local state sets is not more than 7, it is determined that the target user is in a yawn state.

[0055] In order to consider that the first local state may have a certain deviation in recognition, the local part corresponding to the determined user image of the first local state is identified again, which is beneficial to determine the user state of the target user, and the user state of the target user is determined according to the mixed state length, the first state proportion value, the second state proportion value, and the plurality of first local state sets, including: The local part of the user image containing the first local state is subjected to feature extraction to obtain a local feature; The length pixel number and the width pixel number corresponding to the local feature are determined, the length pixel number is used to represent the number of pixels occupied by the local feature in the length direction, and the width pixel number is used to represent the number of pixels occupied by the local feature in the width direction; The ratio between the length pixel number and the width pixel number is determined to obtain a local feature ratio; The user state of the target user is determined according to the local feature ratio, the mixed state length, the first state proportion value, the second state proportion value, and the plurality of first state sets in the user image set.

[0056] The local feature refers to the feature data of the local part, such as mouth feature data, eye feature data, etc.

[0057] The local feature ratio refers to the ratio of the pixels occupied by the local feature along the length direction and the width direction, respectively. For example, the local feature is composed of 5*8 pixels, the length pixel number is 5, and the width pixel number is 8.

[0058] For example, when the local part is the mouth, the preset condition provides that the local feature ratio is greater than 5 / 7, and the user local state is recognized as open mouth. The provision of the preset condition is not limited here.

[0059] For example, the preset condition provides that when the mixed state length is not less than 15, the first proportion value is not less than 0.3, the second proportion value is not more than 0.6, the number of the plurality of first local state sets is not more than 7, the local feature ratio is greater than 5 / 7, and the number of frames in which the local feature ratio is greater than 5 / 7 is greater than 7, it is determined that the target user is in a yawn state.

[0060] In order to consider how to determine the user state according to the continuous state change result, the user local state includes a first local state, and the user state in which the target user is determined to be in according to the continuous state change result includes: According to the continuous state change result, a group of continuous user images with the first local state in the user image set is obtained to obtain a first local state set; A proportion of the first local state set in the user image set is determined to obtain a proportion value; The user state in which the target user is determined to be in according to the proportion value.

[0061] The first local state set is composed of a plurality of continuous user images, and each user image has the same local state. For example, the first local state set is composed of a plurality of continuous user images with wide-open mouth, and can also be composed of a plurality of continuous user images with small-open eyes or closed eyes, etc.

[0062] The proportion value is used to represent the proportion of the first local state set in the user image set.

[0063] For example, if it is determined whether the target user yawns, the proportion of a group of continuous user images with wide-open mouth in the user image set is determined, and when the proportion value corresponding to the proportion satisfies the preset condition (the proportion value is not less than 10%), it is determined that the target user is in a yawn state.

[0064] For example, if it is determined whether the target user has eye abnormalities, the proportion of a group of continuous user images with closed eyes or small-open eyes in the user image set is determined, and when the proportion value corresponding to the proportion satisfies the preset condition (the proportion value is not less than 10%), it is determined that the target user is in an eye abnormal state.

[0065] For example, when the target user is a driver, according to the individual differences of the user, some users have small-open eyes when sleeping, so when the proportion of a group of continuous user images with closed eyes or small-open eyes in the user image set satisfies the preset condition, it is determined that the target user is in an eye abnormal state.

[0066] In this embodiment, the method of the algorithm will be described in detail by taking the recognition of the target user yawn as an example, and the specific process is as follows: Obtaining a user image set of the target user in a preset time period, the user image set comprising a plurality of user images.

[0067] Performing face detection on the user image, intercepting a largest face in the user image as a face of the target user, and obtaining a target user face.

[0068] Performing state recognition on a mouth of the target user face, and obtaining a user local state corresponding to the user image.

[0069] In some embodiments, the user local state comprises closed mouth, small mouth opening, and wide mouth opening.

[0070] According to an image acquisition sequence, sequentially performing state change recognition on the user local state corresponding to the user image in the user image set, and obtaining a continuous state change result.

[0071] In some embodiments, sequentially performing state change recognition on the user local state corresponding to the user image in the user image set, and obtaining a continuous state change result, comprises: Obtaining mapping relationship information, the mapping relationship information comprising a mapping relationship between the user local state and a state identifier; According to the mapping relationship information, determining a state identifier associated with the user local state corresponding to the user image, and obtaining identifier information; According to the user image set and the identifier information, obtaining an identifier information set; According to a preset rule and an image acquisition sequence, constructing a dynamic change graph corresponding to the identifier information in the identifier information set, and obtaining the continuous state change result.

[0072] For example, the mapping relationship information records that the closed mouth, the small mouth opening, and the wide mouth opening correspond to the numbers 0, 1, and 2 respectively, and after obtaining the user local state corresponding to the user image, the user local state corresponding to the user image set is converted into a sequence of 0 / 1 / 2 numbers.

[0073] According to the continuous state change result, determining a user state in which the target user is located.

[0074] In some embodiments, the user state at least comprises a first state, and according to the continuous state change result, determining the user state in which the target user is located, comprises: When the continuous state change result satisfies a preset condition, determining the user state of the target user as the first state.

[0075] In some embodiments, when the continuous state change result is a waveform graph, the closed mouth is a wave trough, the wide mouth opening is a wave peak, and the small mouth opening is between the wave trough and the wave peak, and when the continuous state change result satisfies a preset condition, determining the user state of the target user as the first state, comprises: When the proportion of user images containing continuous open-mouth images in the peak in the user image set is not less than 10%, the user state of the target user is determined as the first state.

[0076] In some embodiments, when the continuous state change result is a waveform graph, the closed mouth is a trough, the open mouth is a peak, the slightly open mouth is between the trough and the peak, and when the continuous state change result meets the preset condition, the user state of the target user is determined as the first state, including: Obtaining a set of continuous user images with a first local state and a second local state, obtaining a mixed state set; Determining the length of the mixed state set in the user image set, obtaining a mixed state length; Determining the proportion of user images corresponding to the first local state and the second local state in the user image set, obtaining a first proportion value and a second proportion value; Obtaining a plurality of first local state sets from the user image set, each of which is a set of continuous user images with the first local state; According to the mixed state length, the first proportion value, the second proportion value, and the plurality of first local state sets, determining the user state in which the target user is located.

[0077] For example: Rule 1: The maximum length of the mixed state set containing consecutive substrings of 1 and 2 (not including 0) is not less than 15, and there is a certain length of open-mouth state.

[0078] Rule 2: The second proportion value is the proportion of slightly open mouth, which is not more than 0.6, excluding always open mouth or part of speaking and eating.

[0079] Rule 3: The number of peaks (open mouth) in the plurality of first local state sets is not more than 7, excluding the case of always open mouth or closed mouth during speaking and eating, and the peak number is counted according to the following rule: the same continuous state is merged into one state, the number of 2 is counted, for example, 00000011112222222111111 will become 0121, and the peak number is 1.

[0080] Rule 4: The first proportion value is the proportion of open mouth, which is not less than 0.3, and there is a certain length of open-mouth state.

[0081] In some embodiments, according to the mixed state length, the first state proportion value, the second state proportion value, and the plurality of first local state sets, the user state in which the target user is located is determined, including: Extracting local features from the user images containing the first local state, obtaining local features; Determine the length pixel number and the width pixel number of the local feature, the length pixel number is used to represent the pixel number of the local feature in the length direction, and the width pixel number is used to represent the pixel number of the local feature in the width direction; Determine the ratio between the length pixel number and the width pixel number, and obtain the local feature ratio; According to the local feature ratio, the mixed state length, the first state proportion ratio, the second state proportion ratio and the plurality of first state sets in the user image set, determine the user state in which the target user is located.

[0082] Rule 5: The local feature ratio is the ratio between the mouth width pixel number and the length pixel number, which is greater than 1.4, and the user image in which the local feature ratio in the user image set is greater than 1.4 is greater than 7, further confirming the certain length of the open mouth state.

[0083] As can be seen from the above, the technical scheme can obtain a user image set of a target user in a preset time period, and the user image set includes a plurality of user images; the state of a local part of the face of the target user in the user image is recognized to obtain a user local state corresponding to the user image; the state change of the user local state corresponding to the user image in the user image set is recognized in sequence according to the image acquisition sequence to obtain a continuous state change result, and the image acquisition sequence is the sequence corresponding to the acquisition of the user image; and the user state in which the target user is located is determined according to the continuous state change result.

[0084] The above only describes the preferred examples of the present application and is not used to limit the present application. Although the present application is described in detail with reference to the foregoing examples, those skilled in the art can still modify the technical solutions recorded in the foregoing examples or make equivalent replacement of some technical features. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for identifying and evaluating the real-time status of a vehicle driver, characterized in that: The specific steps include: Step S110: obtaining a user image set of a target user within a preset time period, where the user image set includes multiple user images; Step S120: performing state recognition on a local part of the target user's face in the user image to obtain the user's local state corresponding to the user image; Step S130: performing state change recognition on the user local states corresponding to the user images in the user image set in sequence according to the image acquisition order, and obtaining a continuous state change result. The image acquisition order is the order corresponding to the acquisition of the user images; Step S140: Determine the user state of the target user according to the continuous state change result, and finally obtain the identification evaluation result.

2. A method for identifying and evaluating a vehicle driver's real-time state according to claim 1, characterized in that: The user state includes at least a first state, and determining the user state of the target user according to the continuous state change result includes: when the continuous state change result meets a preset condition, determining the user state of the target user as the first state.

3. The method for real-time identification and evaluation of a vehicle driver's state according to claim 1, characterized in that: The state change identification in step S130 specifically includes the following steps: Step S131: Acquire mapping relationship information, where the mapping relationship information includes a mapping relationship between a user's local state and a state identifier; Step S132: determining the state identifier associated with the user local state corresponding to the user image based on the mapping relationship information, and obtaining identifier information; Step S133: Obtaining an identification information set based on the user image set and identification information; Step S134: constructing a dynamic change graph corresponding to the identification information in the identification information set according to the preset rules and the image acquisition sequence to obtain a continuous state change result.

4. The method for real-time identification and evaluation of a vehicle driver's state according to claim 3, characterized in that: The step S134 of constructing a dynamic change graph corresponding to the identification information in the identification information set includes the following specific steps: Step S1341: merging multiple consecutive and identical identification information in the identification information set into one identification information to obtain new identification information; Step S1342: Obtain a new identification information set based on the new identification information and the identification information set; Step S1343: constructing a dynamic change graph corresponding to the new identification information in the new identification information set according to the preset rules and the image acquisition sequence, and obtaining a continuous state change result.

5. The method for real-time identification and evaluation of a vehicle driver's state according to claim 1, characterized in that: The user partial state of step S120 includes a first partial state and a second partial state. The step S140 of determining the user state of the target user includes the following specific steps: Step S1401: Acquire a set of continuous user images having a first partial state and a second partial state to obtain a mixed state set; Step S1402: Determine the length of the mixed state set in the user image set to obtain the mixed state length; Step S1403: determining the proportions of user images corresponding to the first partial state and the second partial state in the user image set, respectively, to obtain a first proportion value and a second proportion value; Step S1404: acquiring multiple sets of continuous user images having a first partial state from the user image set, to obtain multiple first partial state sets; Step S1405: Determine the user state of the target user according to the mixed state length, the first proportion value, the second proportion value, and the plurality of first partial state sets.

6. The method for real-time identification and evaluation of a vehicle driver's state according to claim 5, characterized in that: The step S1405 of determining the user state of the target user according to the mixed state length, the first proportion value, the second proportion value, and the plurality of first partial state sets comprises the following specific steps: Step S14051: extracting features of local parts of the user image containing the first local state to obtain local features; Step S14052: determining the number of length pixels and width pixels corresponding to the local feature, where the number of length pixels is used to represent the number of pixels occupied by the local feature in the length direction, and the number of width pixels is used to represent the number of pixels occupied by the local feature in the width direction; Step S14053: determining the ratio between the number of length pixels and the number of width pixels to obtain a local feature ratio; Step S14054: Determine the user state of the target user based on the local feature ratio, mixed state length, first state proportion, second state proportion, and multiple first state sets in the user image set.