Face fatigue monitoring and distraction behavior analysis system based on YOLOv8
Through the YOLOv8-based facial fatigue monitoring and distracted behavior analysis system, the problems of monitoring lag and insufficient accuracy in the prior art are solved, real-time and accurate monitoring of the fatigue status and distracted behavior of the driver or operator are achieved, and the improvement of the Monitor accuracy and reaction speed.
Patent Information
- Application Number
- CN202411568071.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-05
- Publication Date
- 2025-02-18
AI Technical Summary
When monitoring the fatigue status and distracting behavior of the driver or operator, the prior art has problems of monitoring lag and insufficient accuracy, and it is impossible to capture abnormal states in real time and accurately.
Using a facial fatigue monitoring and distraction behavior analysis system based on YOLOv8, images were collected through high-definition cameras, and fatigue state recognition model and distraction behavior recognition model were constructed based on the optimized YOLOv8 model to identify fatigue and distraction situations in real time.
Real-time and accurate monitoring of the fatigue status and distracted behavior of the driver or operator is achieved, monitoring accuracy and reaction speed are improved, and more reliable safety guarantees are provided.
Smart Images
Figure CN119479037A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of security monitoring technology, and more specifically, to a facial fatigue monitoring and distraction behavior analysis system based on YOLOv8. Background Art
[0002] In today's rapidly developing modern transportation and work safety-related fields, fatigue driving, distraction and other distracting behaviors have become key factors in causing traffic accidents and work accidents.
[0003] Although the current monitoring methods can obtain relevant data, there are problems with monitoring lag and insufficient accuracy, and they are unable to capture the abnormal status of drivers or other operators in the working environment in real time and accurately.
[0004] In order to solve the above problems, a technical solution is now provided. Summary of the invention
[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a facial fatigue monitoring and distraction behavior analysis system based on YOLOv8, which collects images through a high-definition camera, builds a fatigue state recognition model and a distraction behavior recognition model based on an optimized YOLOv8 model, and accurately recognizes the fatigue state and distraction behavior of the driver or operator in real time, providing reliable technical support for improving driving safety and work efficiency, so as to solve the problems raised in the above-mentioned background technology.
[0006] To achieve the above object, the present invention provides the following technical solutions:
[0007] A facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 monitors and analyzes the facial expressions and behaviors of a driver or other working environment in real time, and identifies fatigue and distraction. The facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 comprises an acquisition module, a marking module, a model building module, a recognition module and a data recording and alarm module; the acquisition module acquires the facial image, head image and hand image of the driver or operator in real time; the marking module pre-processes the acquired image and marks the key parts in the image; the model building module is used to build a fatigue state recognition model and an analysis behavior recognition model based on an optimized YOLOv8 model; the recognition module performs real-time fatigue behavior and analysis behavior recognition and analysis on the acquired image data; the data recording and alarm module records, stores, analyzes and responds to the recognized fatigue behavior and distraction behavior in real time.
[0008] As a further solution of the present invention, the acquisition module acquires the facial image, head image and hand image of the driver or operator in real time, including the following specific contents:
[0009] The acquisition module captures the driver or operator's facial image, head image and hand image in real time through a high-definition camera. The facial image mainly focuses on the eyes, mouth and nose areas, the head image focuses on the head posture and head movement trajectory, and the hand image focuses on the hand movement and position. In order to ensure the real-time and accuracy of the image acquisition module, the acquisition module uses a 12-megapixel high-definition camera and a high-speed frame rate of 30 frames per second. Since the acquired images often have problems such as noise, blur and uneven lighting, the acquisition module will improve the image clarity through denoising algorithms and image enhancement technologies. Under low-light conditions, automatic gain control and automatic white balance functions are used to ensure the brightness and color balance of the image. At the same time, geometric transformation technology is used to correct the image and adjust the distortion problem to improve the image quality. The acquisition module crops and scales the acquired image, processes the image to a size of 512x512 pixels, and prepares for inputting the acquired image into the YOLOv8 model.
[0010] As a further solution of the present invention, the marking module pre-processes the collected image and marks the key parts in the image, including the following specific contents:
[0011] The marking module processes the collected facial image and identifies the main facial feature areas, which include the key parts of the eyes, mouth and nose. After identifying the main facial areas, the marking module will accurately mark the facial contour and generate a contour annotation image of the facial area; the marking module processes the collected head image, extracts the key features of the head, and identifies the posture changes of the head by marking the head contour. The posture changes of the head include drooping the head, frequent shaking, leaning to a certain direction, etc. The marking module will segment the head area and extract key points through a deep learning network to obtain the posture information of the head; the marking module calibrates the contour area of the hand by collecting the hand image.
[0012] As a further solution of the present invention, the model building module is used to build a fatigue state recognition model and an analysis behavior recognition model based on the optimized YOLOv8 model, including the following specific contents:
[0013] The YOLOv8 model is composed of three different modules, including Backbone, Neck and Head. The Backbone is the main feature extraction part, which is used to extract the deep features of the input image. In the Backbone, the input image passes through multiple layers of Conv, C2f layers and an SPPF module. The SPPF module splices the extraction results of different feature layers through three maximum pooling layers, reducing the computational complexity; then, the fused feature information is passed to Neck for further processing; the Neck integrates the pyramid attention network and the feature pyramid network to enhance the network's ability to extract features. The network architecture that integrates the pyramid attention network and the feature pyramid network helps the YOLOv8 model to fully acquire information at multiple scales, allowing the network to better acquire features at different scales; the Head adopts a decoupled head structure, uses an independent dual-branch strategy for target classification and bounding box regression, and uses different loss functions to optimize the learning process, thereby improving recognition accuracy;
[0014] In order to extract more feature information, the channel attention mechanism (Squeeze-and-ExcitationAttention, SEAttention) is integrated into the Backbone. The channel attention mechanism adaptively adjusts the importance of each channel by learning weights, so that the YOLOv8 model can pay more attention to the important features in the input information. The working steps of the channel attention mechanism include:
[0015] Using global average pooling technology, the features and information of each channel are collected, so that the YOLOv8 model can grasp the global information on each channel, so as to better understand the distribution of the overall features;
[0016] Two fully connected layers are used. The purpose of the two fully connected layers is to first reduce the dimension and then project the result back to the original number of channels. Reducing the complexity of the model can also improve the generalization performance of the model. The two fully connected layers grasp the weight value of each channel, allowing the YOLOv8 model to flexibly strengthen or suppress the features in each channel.
[0017] Multiply the learned attention weights with the original input features to scale the features of each channel, strengthening key features and suppressing unimportant features;
[0018] A fatigue state recognition model and an analysis behavior recognition model are constructed according to the optimized YOLOv8 model to recognize fatigue behavior and analysis behavior.
[0019] As a further solution of the present invention, the recognition module performs real-time fatigue behavior and analysis behavior recognition and analysis on the collected image data, including the following specific contents:
[0020] The recognition module identifies the fatigue behavior and distraction behavior of the driver or operator based on the fatigue state recognition model and analysis behavior recognition model constructed by the YOLOv8 model; by identifying the opening and closing of the eyes and the frequent blinking behavior, it can be judged whether the driver or operator is at risk of fatigue. Frequent or long-term closure of the eyes will be marked as a fatigue state, and the opening and closing degree of the mouth and the yawning action are recognized as the basis for fatigue assessment. In addition, the recognition module can recognize that the driver or operator's head is drooping, frequently shaking or deviating from the normal direction, indicating that the driver or operator has entered a fatigue state; in addition to monitoring the driver's or operator's face and head, the recognition module will also analyze the driver's or operator's hands and sight direction. When the driver or operator's eyes are not focused on the front, or there is a distraction in operation, the distraction behavior includes looking down at the mobile phone and smoking, the recognition module determines and outputs the distraction behavior result; in order to ensure that the recognition module can be monitored in real time in a high-speed vehicle or in a working environment, the recognition module can complete the processing of each frame of the image within milliseconds.
[0021] As a further solution of the present invention, the data recording and alarm module records, stores, analyzes and responds to the detected fatigue and distraction behaviors in real time, including the following specific contents:
[0022] After the driver or staff's fatigue or distraction behavior is accurately detected by the recognition module, the data recording and alarm module will immediately record the event and save the relevant information to the database; the recorded content includes the time when the behavior occurred, the type of behavior, the duration of the behavior, and the frequency of occurrence. The data recording and alarm module will also automatically store the environmental data when the event occurred, and the environmental data includes environmental image data and vehicle speed data; the data recording and alarm module calculates the frequency of fatigue or distraction behavior in different time periods, duration, high-risk time period, and type of behavior through aggregate analysis of behavioral data. Through statistical results, managers can identify potential risk points of fatigue or distraction behavior and take corresponding preventive measures; the data recording and alarm module can immediately respond and alarm when fatigue or distraction behavior is detected. The types of alarms include sound alarms, vibration reminders and screen display warnings. The sound alarm uses a warning sound with adjustable volume, such as repeated beeps, to ensure that the warning can be clearly heard in a noisy environment; the vibration reminder is suitable for vehicle-mounted systems or wearable devices, and can remind the driver or operator to pay attention to safety through vibration; the data recording and alarm module also has the function of adjusting the alarm intensity according to different levels of fatigue and distraction behavior. When the recognition module detects a slight fatigue state of the driver or operator, the slight fatigue state includes: brief eye closure or slight head movement, the data recording and alarm module emits a slight prompt sound or vibration to remind the driver or operator to pay attention. When the recognition module detects severe fatigue behavior or high-risk distraction behavior, the alarm intensity will be automatically adjusted to a high-intensity warning, such as a loud alarm or continuous vibration. The severe fatigue behavior includes: eyes closed for a long time, yawning multiple times, and head continuously drooping. The high-risk distraction behavior includes: using a mobile phone.
[0023] As a further solution of the present invention, the facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 can be integrated into multiple platforms. The facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 can be deployed not only on traditional PC or server sides, but also on vehicle-mounted platforms, mobile devices and embedded platforms. With the diversification of application scenarios, the facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 also supports integration with heart rate monitoring sensors and environmental monitoring sensors to achieve more comprehensive fatigue and distraction behavior monitoring.
[0024] As a further solution of the present invention, a facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 has a friendly user interface, provides a simple and easy-to-use user interface, and displays monitoring data and alarm information in real time; users can customize parameters such as alarm thresholds and monitoring frequencies as needed to improve the flexibility and adaptability of the system.
[0025] The technical effects and advantages of the facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 of the present invention are as follows:
[0026] The present invention collects and processes images through a high-definition camera, a marking module preprocesses images and marks key parts, a model building module optimizes the YOLOv8 model to build a fatigue and distraction behavior recognition model, the recognition module accurately recognizes fatigue and distraction behaviors in real time, and the data recording and alarm module records, stores and analyzes data and issues multiple alarms according to the behavior level. It can integrate multiple platforms and support integration with multiple sensors. It has a friendly user interface and can customize parameters, effectively improves monitoring accuracy and response speed, monitors fatigue and distraction behaviors in real time, provides efficient and reliable protection for fields such as transportation and work safety, helps managers take preventive measures to reduce accident risks, and improves system flexibility and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] Figure 1 This is a structural schematic diagram of a facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 in the present invention.
[0028] Figure 2 This is a schematic diagram of the YOLOv8 model structure after optimization in the present invention.
[0029] Figure 3 This is a diagram showing the monitoring effect of facial fatigue monitoring and distraction behavior analysis based on YOLOv8. DETAILED DESCRIPTION
[0030] 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.
[0031] Example 1
[0032] See also Figure 1As shown in the structural schematic diagram, an embodiment of the present invention provides a facial fatigue monitoring and distraction behavior analysis system based on YOLOv8, including an acquisition module, a marking module, a model building module, a recognition module and a data recording and alarm module; the acquisition module acquires the facial image, head image and hand image of the driver or operator in real time; the marking module preprocesses the acquired image and marks the key parts in the image; the model building module is used to build a fatigue state recognition model and an analysis behavior recognition model based on the optimized YOLOv8 model; the recognition module performs real-time fatigue behavior and analysis behavior recognition and analysis on the acquired image data; the data recording and alarm module records, stores, analyzes and responds to the alarm for the recognized fatigue behavior and distraction behavior in real time.
[0033] Furthermore, the acquisition module acquires the facial image, head image and hand image of the driver or staff in real time, including the following specific contents:
[0034] The acquisition module captures the driver or operator's facial image, head image and hand image in real time through a high-definition camera. The facial image mainly focuses on the eyes, mouth and nose areas, the head image focuses on the head posture and head movement trajectory, and the hand image focuses on the hand movement and position. In order to ensure the real-time and accuracy of the image acquisition module, the acquisition module uses a 12-megapixel high-definition camera and a high-speed frame rate of 30 frames per second. Since the acquired images often have problems such as noise, blur and uneven lighting, the acquisition module will improve the image clarity through denoising algorithms and image enhancement technologies. Under low-light conditions, automatic gain control and automatic white balance functions are used to ensure the brightness and color balance of the image. At the same time, geometric transformation technology is used to correct the image and adjust the distortion problem to improve the image quality. The acquisition module crops and scales the acquired image, processes the image to a size of 512x512 pixels, and prepares for inputting the acquired image into the YOLOv8 model.
[0035] Furthermore, the marking module pre-processes the collected images and marks the key parts in the images, including the following specific contents:
[0036] The marking module processes the collected facial image and identifies the main facial feature areas, which include the key parts of the eyes, mouth and nose. After identifying the main facial areas, the marking module will accurately mark the facial contour and generate a contour annotation image of the facial area; the marking module processes the collected head image, extracts the key features of the head, and identifies the posture changes of the head by marking the head contour. The posture changes of the head include drooping the head, frequent shaking, leaning to a certain direction, etc. The marking module will segment the head area and extract key points through a deep learning network to obtain the posture information of the head; the marking module calibrates the contour area of the hand by collecting the hand image.
[0037] See also Figure 2 As shown in the structural schematic diagram, the embodiment of the present invention provides an optimized YOLOv8 model, and the model construction module is used to construct a fatigue state recognition model and an analysis behavior recognition model based on the optimized YOLOv8 model, including the following specific contents:
[0038] The YOLOv8 model is composed of three different modules, including Backbone, Neck and Head. The Backbone is the main feature extraction part, which is used to extract the deep features of the input image. In the Backbone, the input image passes through multiple layers of Conv, C2f layers and an SPPF module. The SPPF module splices the extraction results of different feature layers through three maximum pooling layers, reducing the computational complexity; then, the fused feature information is passed to Neck for further processing; the Neck integrates the pyramid attention network and the feature pyramid network to enhance the network's ability to extract features. The network architecture that integrates the pyramid attention network and the feature pyramid network helps the YOLOv8 model to fully acquire information at multiple scales, allowing the network to better acquire features at different scales; the Head adopts a decoupled head structure, uses an independent dual-branch strategy for target classification and bounding box regression, and uses different loss functions to optimize the learning process, thereby improving recognition accuracy;
[0039] In order to extract more feature information, the channel attention mechanism (Squeeze-and-ExcitationAttention, SEAttention) is integrated into the Backbone. The channel attention mechanism adaptively adjusts the importance of each channel by learning weights, so that the YOLOv8 model can pay more attention to the important features in the input information. The working steps of the channel attention mechanism include:
[0040] Using global average pooling technology, the features and information of each channel are collected, so that the YOLOv8 model can grasp the global information on each channel, so as to better understand the distribution of the overall features;
[0041] Two fully connected layers are used. The purpose of the two fully connected layers is to first reduce the dimension and then project the result back to the original number of channels. Reducing the complexity of the model can also improve the generalization performance of the model. The two fully connected layers grasp the weight value of each channel, allowing the YOLOv8 model to flexibly strengthen or suppress the features in each channel.
[0042] Multiply the learned attention weights with the original input features to scale the features of each channel, strengthening key features and suppressing unimportant features;
[0043] A fatigue state recognition model and an analysis behavior recognition model are constructed according to the optimized YOLOv8 model to recognize fatigue behavior and analysis behavior.
[0044] Further, see Figure 3 As shown in the effect display diagram, the recognition module performs real-time fatigue behavior and analysis behavior recognition and analysis on the collected image data, including the following specific contents:
[0045] The recognition module identifies the fatigue behavior and distraction behavior of the driver or operator based on the fatigue state recognition model and analysis behavior recognition model constructed by the YOLOv8 model; by identifying the opening and closing of the eyes and the frequent blinking behavior, it can be judged whether the driver or operator is at risk of fatigue. Frequent or long-term closure of the eyes will be marked as a fatigue state, and the opening and closing degree of the mouth and the yawning action are recognized as the basis for fatigue assessment. In addition, the recognition module can recognize that the driver or operator's head is drooping, frequently shaking or deviating from the normal direction, indicating that the driver or operator has entered a fatigue state; in addition to monitoring the driver's or operator's face and head, the recognition module will also analyze the driver's or operator's hands and sight direction. When the driver or operator's eyes are not focused on the front, or there is a distraction in operation, the distraction behavior includes looking down at the mobile phone and smoking, the recognition module determines and outputs the distraction behavior result; in order to ensure that the recognition module can be monitored in real time in a high-speed vehicle or in a working environment, the recognition module can complete the processing of each frame of the image within milliseconds.
[0046] Furthermore, the data recording and alarm module records, stores, analyzes and responds to detected fatigue and distraction behaviors in real time, including the following specific contents:
[0047] After the driver or staff's fatigue or distraction behavior is accurately detected by the recognition module, the data recording and alarm module will immediately record the event and save the relevant information to the database; the recorded content includes the time when the behavior occurred, the type of behavior, the duration of the behavior, and the frequency of occurrence. The data recording and alarm module will also automatically store the environmental data when the event occurred, and the environmental data includes environmental image data and vehicle speed data; the data recording and alarm module calculates the frequency of fatigue or distraction behavior in different time periods, duration, high-risk time period, and type of behavior through aggregate analysis of behavioral data. Through statistical results, managers can identify potential risk points of fatigue or distraction behavior and take corresponding preventive measures; the data recording and alarm module can immediately respond and alarm when fatigue or distraction behavior is detected. The types of alarms include sound alarms, vibration reminders and screen display warnings. The sound alarm uses a warning sound with adjustable volume, such as repeated beeps, to ensure that the warning can be clearly heard in a noisy environment; the vibration reminder is suitable for vehicle-mounted systems or wearable devices, and can remind the driver or operator to pay attention to safety through vibration; the data recording and alarm module also has the function of adjusting the alarm intensity according to different levels of fatigue and distraction behavior. When the recognition module detects a slight fatigue state of the driver or operator, the slight fatigue state includes: brief eye closure or slight head movement, the data recording and alarm module emits a slight prompt sound or vibration to remind the driver or operator to pay attention. When the recognition module detects severe fatigue behavior or high-risk distraction behavior, the alarm intensity will be automatically adjusted to a high-intensity warning, such as a loud alarm or continuous vibration. The severe fatigue behavior includes: eyes closed for a long time, yawning multiple times, and head continuously drooping. The high-risk distraction behavior includes: using a mobile phone.
[0048] Furthermore, the facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 can be integrated into multiple platforms. The facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 can be deployed not only on traditional PC or server sides, but also on vehicle-mounted platforms, mobile devices and embedded platforms. With the diversification of application scenarios, the facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 also supports integration with heart rate monitoring sensors and environmental monitoring sensors to achieve more comprehensive fatigue and distraction behavior monitoring.
[0049] Furthermore, the facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 has a friendly user interface, providing a simple and easy-to-use user interface, and displaying monitoring data and alarm information in real time; users can customize parameters such as alarm thresholds and monitoring frequencies as needed to improve the flexibility and adaptability of the system.
[0050] The present invention collects and processes images through a high-definition camera, a marking module preprocesses images and marks key parts, a model building module optimizes the YOLOv8 model to build a fatigue and distraction behavior recognition model, the recognition module accurately recognizes fatigue and distraction behaviors in real time, and the data recording and alarm module records, stores and analyzes data and issues multiple alarms according to the behavior level. It can integrate multiple platforms and support integration with multiple sensors. It has a friendly user interface and can customize parameters, effectively improves monitoring accuracy and response speed, monitors fatigue and distraction behaviors in real time, provides efficient and reliable protection for fields such as transportation and work safety, helps managers take preventive measures to reduce accident risks, and improves system flexibility and adaptability.
[0051] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
[0052] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A facial fatigue monitoring and distraction behavior analysis system based on YOLOv8, which monitors and analyzes the facial expressions and behaviors of drivers or other working environments in real time, identifies fatigue and distraction, and is characterized by: The facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 includes an acquisition module, a marking module, a model building module, a recognition module and a data recording and alarm module; the acquisition module acquires the facial image, head image and hand image of the driver or operator in real time; the marking module preprocesses the acquired images and marks the key parts in the images; the model building module is used to build a fatigue state recognition model and an analysis behavior recognition model based on the optimized YOLOv8 model; the recognition module performs real-time fatigue behavior and analysis behavior recognition and analysis on the acquired image data; the data recording and alarm module records, stores, analyzes and responds to alarms for the recognized fatigue behavior and distraction behavior in real time.
2. A facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 according to claim 1, characterized in that ,The model building module is used to build a fatigue state recognition model and ,analysis behavior recognition model based on the optimized YOLOv8 model, including the following contents: The YOLOv8 model consists of three different modules, including Backbone, Neck and Head. The Backbone is the main feature extraction part, which is used to extract the deep features of the input image. In order to extract more feature information, the channel attention mechanism is integrated into the Backbone. The channel attention mechanism adaptively adjusts the importance of each channel by learning weights, so that the YOLOv8 model can pay more attention to the important features in the input information. The working steps of the channel attention mechanism include: Using global average pooling technology, the features and information of each channel are collected, so that the YOLOv8 model can grasp the global information on each channel, so as to better understand the distribution of the overall features; Two fully connected layers are used. The purpose of the two fully connected layers is to first reduce the dimension and then project the result back to the original number of channels. Reducing the complexity of the model can also improve the generalization performance of the model. The two fully connected layers grasp the weight value of each channel, allowing the YOLOv8 model to flexibly strengthen or suppress the features in each channel. Multiply the learned attention weights with the original input features to scale the features of each channel, strengthening key features and suppressing unimportant features; A fatigue state recognition model and an analysis behavior recognition model are constructed according to the optimized YOLOv8 model to recognize fatigue behavior and analysis behavior.
3. A facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 according to claim 1, characterized in that The acquisition module captures the driver's or operator's facial image, head image and hand image in real time through a 12-megapixel high-definition camera and a high-speed frame rate of 30 frames per second. The facial image focuses on the driver's or operator's eyes, mouth and nose areas, the head image focuses on the driver's or operator's head posture and head movement trajectory, and the hand image focuses on the driver's or operator's hand movements and positions.
4. A facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 according to claim 3, characterized in that ,The acquisition module uses denoising algorithm and image enhancement ,technology to improve image clarity, uses automatic gain control and automatic ,white balance functions to ensure image brightness and color balance under low light conditions, uses ,geometric transformation technology to correct image distortion problems, and ,crops and scales the image to 512x512 pixel size.
5. A facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 according to claim 1, characterized in that The marking module identifies the key parts of the eyes, mouth and nose in the facial image, and accurately marks the facial contour to generate a contour annotation image of the facial area; the marking module extracts the key features of the head from the head image, and identifies the changes in head posture by marking the head contour, and the changes in head posture include lowering the head and frequent shaking; the marking module calibrates the hand contour area of the collected hand image.
6. A facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 according to claim 1, characterized in that The recognition module builds a fatigue state recognition model and an analysis behavior recognition model based on the YOLOv8 model. It judges fatigue and distraction behaviors by detecting the driver or operator's eye opening and closing, blinking, mouth opening and closing, yawning, head posture, and hand and sight direction, and can process each frame of the image at the millisecond level.
7. A facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 according to claim 1, characterized in that After detecting driver or operator fatigue or distraction, the data recording and alarm module immediately records the event and saves relevant information to the database. The recorded content includes the time, type, duration and frequency of the behavior. The data recording and alarm module will also automatically store the environmental data when the event occurs. The environmental data includes environmental image data and vehicle speed data. The data recording and alarm module calculates the frequency, duration, high-risk time period and type of driver or operator fatigue or distraction in different time periods through aggregate analysis of the behavior data.
8. The facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 according to claim 1, characterized in that The data recording and alarm module can immediately respond and alarm when fatigue or distraction is detected; the types of alarms include sound alarms, vibration reminders and screen display warnings. The sound alarm uses a warning sound with adjustable volume, and the vibration reminder is suitable for vehicle-mounted systems or wearable devices, and can remind drivers or operators to pay attention to safety through vibration.
9. The facial fatigue monitoring and distraction behavior analysis system based on YOLOv8 according to claim 1, characterized in that ,The data recording and alarm module adjusts the alarm intensity according to different ,levels of fatigue and distraction behaviors, emitting a slight ,prompt sound or vibration for a slight fatigue state, and a high-intensity warning for severe ,fatigue behavior or high-risk distraction behavior.