Driver Attention Research Articles

Analysis and Prediction of Risky Driving Behaviors Using Fuzzy Analytical Hierarchy Process and Machine Learning Techniques

Driver behavior plays a pivotal role in ensuring road safety as it is a significant factor in preventing traffic crashes. Although extensive research has been conducted on this topic in developed countries, there is a notable gap in understanding driver behavior in developing countries, such as Pakistan. It is essential to recognize that the cultural nuances, law enforcement practices, and government investments in traffic safety in Pakistan are significantly different from those in other regions. Recognizing this disparity, this study aims to comprehensively understand risky driving behaviors in Peshawar, Pakistan. To achieve this goal, a Driver Behavior Questionnaire was designed, and responses were collected using Google Forms, resulting in 306 valid responses. The study employs a Fuzzy Analytical Hierarchy Process framework to evaluate driver behavior’s ranking criteria and weight factors. This framework assigns relative weights to different criteria and captures the uncertainty of driving thought patterns. Additionally, machine learning techniques, including support vector machine, decision tree, Naïve Bayes, Random Forest, and ensemble model, were used to predict driver behavior, enhancing the reliability and accuracy of the predictions. The results showed that the ensemble machine learning approach outperformed others with a prediction accuracy of 0.84. In addition, the findings revealed that the three most significant risky driving attributes were violations, errors, and lapses. Certain factors, such as clear road signage and driver attention, were identified as important factors in improving drivers’ risk perception. This study serves as a benchmark for policymakers, offering valuable insights to formulate effective policies for improving traffic safety.

Unveiling the determinants of injury severities across age groups and time: A deep dive into the unobserved heterogeneity among pedestrian crashes

Pedestrians, particularly susceptible to road traffic crashes, experience varying injury severities influenced by age and time shifts. This research aims to investigate the differences and temporal shifts in factors influencing pedestrian injury severities across different age groups. To achieve this, three random parameters binary logit models with heterogeneity in the means (and variances) were employed. Four years of pedestrian crash data in Hong Kong were utilized in this study. According to United Nations’ definitions of the young and elderly, pedestrians were categorized into three groups: young (under 25 years old), middle-aged (25–65 years old), and elderly (over 65 years old). Initial likelihood ratio tests indicated temporal stability in the young group between 2019 and 2021, with further tests confirming age transferability and overall temporal stability after integrating the three years of young data. The partially constrained temporal stability approach was then developed to further capture the temporal stability of individual variables and simplify model results. Model results identified factors impacting pedestrian injury severities, encompassing pedestrian, driver, vehicle, temporal, and light condition characteristics. Some contributing variables exhibit age-transferability or temporal stability, such as controlled crossing, near controlled crossing, inattentive driver and private car. However, the significance of most contributors varies across age groups and years, with certain factors being age-specific or year-specific. Out-of-sample predictions underscore the cumulative likelihood of fatal or severe injuries with advancing age, and the middle-aged models showed the highest level of temporal stability regarding the risk of injury severity compared to the other two age models. Moreover, middle-aged pedestrians in Hong Kong faced the highest risk of fatal or severe injuries during the first year of the COVID-19 lockdown (2020), but the risk significantly declined for pedestrians of all age groups in the subsequent year. Based on these findings, targeted preventive measures that take into account age differences have been proposed to effectively enhance pedestrian safety.

Experimental road safety study of the actual driver reaction to the street ads using eye tracking, multiple linear regression and decision trees methods

The article describes the results of a naturalistic driving study done in Kuwait performed by 34 participants wearing a mobile eye tracker to monitor the effect of roadside advertisements on user attention. Eye-tracking (fixations) are the main dependent variable and examined as a function of driving/roadside characteristics, particularly billboards, speed, and so forth. The results obtained were analyzed using traditional statistics (ANOVA test and multiple linear regression) and machine learning (decision tree estimation methods). From the results, it was found that road advertisements negatively affect driver attention and thus road safety. How the level of safety varies with the type and size of advertisement was also investigated. As a consequence, all of the estimates revealed that different aspects of advertising have a detrimental impact on drivers' behavior, and that the duration of fixation and the rate of acceleration before viewing an ad are impacted by advertising type and size, respectively.

MODELLING THE FACTORS AFFECTING CRASH OCCURRENCE AND FREQUENCY RESULTED FROM MOBILE PHONE USE WHILE DRIVING: EVIDENCE FROM AL-NAJAF, IRAQ

The recent reports of road traffic accident statistics in Iraq have disclosed a rise in the number of crash injuries resulted from the use of mobile phone while driving. This paper aims to explore the factors contributing to the occurrence and prevalence of such crashes and near crashes in Al-Najaf governorate, Iraq. A representative sample of 417 drivers were interviewed as part of a questionnaire driving survey. Several frequency and modelling analyses were conducted using the IBM SPSS software. The frequency analysis revealed a high use of mobile phones for calling and texting activities while driving. Almost 20% and 55% of the interviewed drivers reported their involvement in a crash or in a near crash because of such use, respectively. Regarding the developed logistic models, the crash involvement sequential regression analysis revealed that factors such gender, education, handheld phoning, calling-answering frequency, and inadequate driving can affect the likelihood of crash occurrence. In contrast, the ordinal logistic near miss models revealed that age, gender, high phone use rate, and improper driving due to such use are influential factors in rising the likelihood of being in multiple near crashes. The analysis results confirm the influence of using phones in distracting the attention of drivers and hence threating their lives; as a result, these findings would be enlightening for agencies and policy makers interested in highway safety.

Area of Interest Tracking Techniques for Driving Scenarios Focusing on Visual Distraction Detection

On-road driving studies are essential for comprehending real-world driver behavior. This study investigates the use of eye-tracking (ET) technology in research on driver behavior and attention during Controlled Driving Studies (CDS). One significant challenge in these studies is accurately detecting when drivers divert their attention from crucial driving tasks. To tackle this issue, we present an improved method for analyzing raw gaze data, using a new algorithm for identifying ID tags called Binarized Area of Interest Tracking (BAIT). This technique improves the detection of incidents where the driver’s eyes are off the road through binarizing frames under different conditions and iteratively recognizing markers. It represents a significant improvement over traditional methods. The study shows that BAIT performs better than other software in identifying a driver’s focus on the windscreen and dashboard with higher accuracy. This study highlights the potential of our method to enhance the analysis of driver attention in real-world conditions, paving the way for future developments for application in naturalistic driving studies.

A Simulator Study on the Driving Impacts of Four Speed-Calming Measures at Unsignalized Intersections

Intersections, especially those without traffic signals, are prone to frequent traffic accidents and disorder. The regulation of vehicle speed is an effective measure that can be taken in order to mitigate accidents and alleviate disorder at unsignalized intersections. This study aims to assess the effectiveness of four different temporary low-cost speed-calming measures: traffic crash barrels, lane markings, intersection markings, and a combination of the three. In this study, we used a driving simulator to design four speed-calming static measures that are typical of intersections; we then statistically analyzed the speed and acceleration data (recorded by the driving simulator) and the driver’s visual data (recorded by an eye tracker). The simulation results demonstrated that different speed-calming measures have varied effects on the attention and visual fixations of drivers; however, they are all effective in reducing the average driving speed at the intersection. The traffic-channelizing lane demonstrated superior performance compared to the other three measures in terms of its cumulative impact on deceleration, driving comfort, and safety. The experimental findings presented herein broaden the range of currently accessible temporary speed-limiting strategies and provide valuable insight into intersections that require low-cost speed management solutions.

Evaluation of eye-catching effect in highway tunnel entrance area based on saccade behavior

Objective This study aimed to investigate the effects of visual attraction in the entrance areas of highway tunnels on drivers’ saccade behavior and driving safety, with the objective of providing insights for tunnel entrance design and driver education. Methods Fifty participants were recruited for the naturalistic driving experiment. Four different visually attractive driving scenarios (baseline, landscape-style architecture, tip slogan, and billboard) were selected. Eye-tracking technology was utilized to record and analyze the scanning behavior of participating drivers. Various metrics, including saccade duration, frequency, amplitude, velocity, and the peak-to-average ratio of saccade velocity (PARSV), were measured and compared across different scenarios. Results The results revealed different patterns of visual scanning dynamics among the four experimental scenarios, reflecting the different levels of visual-cognitive demands and situation awareness of drivers. The visual attraction in the tunnel entrance area resulted in prolonged saccade duration, increased saccade amplitude, reduced saccade frequency, and slower, less stable saccade velocity. Among these, drivers exhibited the most intricate and unstable scanning behavior, accompanied by the lowest level of situational awareness, particularly under the influence of visual attraction of tip slogan. Conclusions The findings highlight the importance of considering visual attraction in tunnel entrance design. Minimizing unnecessary visual attraction can help reduce distractions and maintain driver attention, ultimately enhancing driving safety. Driver education programs should emphasize the significance of adapting to visual attraction at tunnel entrances to promote safer driving practices. Overall, this study contributes valuable insights into the effects of visual attraction on driver scanning behavior, facilitating the development of strategies to improve tunnel design and driver education.

Safety evaluation and prediction of takeover performance in automated driving considering drivers’ cognitive load: A driving simulator study

Automated vehicles alleviate the need for driver attention and control. However, a takeover request (TOR) to the driver remains essential for emergencies and safety–critical scenarios beyond automation's capability. Thus, accessing drivers’ safety performance in various TOR scenarios is crucial for conditionally automated driving (SAE L3). However, TOR safety performance is seldom examined concerning drivers' cognitive load, despite its presumed relevance to TOR scenarios and human–machine interaction. Moreover, the adequacy of the time window preceding a TOR, critical for TOR safety prediction, remains inadequately explored. This study aims to assess safety performance across diverse TOR scenarios in Level 3 conditional automation, incorporating drivers’ cognitive load, and predict TOR safety by considering the time window's impact. A driving simulator experiment gathered eye movement and driving behavior data from 37 recruited participants. Participants were instructed to take control of the vehicle from automated driving within a time budget (TB) of 3 s or 7 s in obstacle avoidance (OA) or lane keeping (LK) scenarios while engaging in non-driving-related tasks (NDRTs). Participants’ subjective cognitive load in the different TOR scenarios was scaled using NASA-TLX. Furthermore, safe TOR performance was predicted utilizing Convolutional Neural Networks (CNNs) across different time window sizes preceding a TOR. The results indicate that: 1) cognitive loads in takeover scenarios ranked from highest to lowest as TB = 3s_OA, TB = 3s_LK, TB = 7s_OA, TB = 7s_LK; and the cognitive loads of the NDRTs ranked from highest to lowest as mistake finding, texting, chatting, monitoring; 2) the takeover safety performance in the four scenarios from lowest to highest was TB = 3s_OA, TB = 3s_LK, TB = 7s_OA, TB = 7s_LK; likewise, the takeover safety performance during the four NDRTs ranked from lowest to highest as mistake finding, monitoring, texting, chatting; 3) the time window size before the TORs significantly affected the prediction performance of the model. A 30-second window was recommended as optimal for predicting takeover safety using the CNN model, achieving an average F1 score of 0.8120 and 81.98 % accuracy. This study's findings enhance our comprehension of driving behavior characteristics during TOR and offer valuable insights for detecting driver states in conditionally automated driving contexts.

VISUAL ATTENTION OF TRAM DRIVERS AS A STEP TOWARDS INCREASING SAFETY IN PUBLIC TRANSPORT: A COMPARATIVE EYE-TRACKING STUDY BETWEEN NOVICE AND EXPERT TRAM DRIVERS

This paper presents an eye-tracking study capturing the dynamics of ambient and focal attention of tram drivers as an application of modern technology in improving their training. Twenty-three experts and twenty-four novices were invited to take part in the eye-tracking study. We explored the visual attention dynamics of tram drivers while they were watching tram-driving simulation videos. The tram driver’s view: The windshields and control panel were defined as areas of interest (AOIs). In line with expectations, experts were more focused than novices and maintained their concentration for a longer time than novice tram drivers. This difference was significant for regions crucial for monitoring the driving environment (the windshields AOIs). Monitoring drivers’ visual attention gives new insights into the process of gaining expertise in scanning the driving environment.

Eye Gaze for Monitoring Attention Through Hybrid Ensemble Learning

One of the countless tasks that call attention to monitoring is necessary for comprising healthcare, education, transportation safety, and human-computer interaction. This research describes novel work done in attention monitoring by fusing a hybrid eye gaze model with deep learning to monitor a driver's attention level. The hybrid eye gaze model proposed is described and its results are produced in this paper. The proposed model uses an augmented dataset where data augmentation techniques like rotation, shifting, shearing, and flipping are applied together with adjustments like changing the fill mode in terms of zooming into the image and rescaling. These are all crucial aspects in reliable and consistent training of the model. Our model is built on modern pre-trained architectures which include VGG16, VGG19, InceptionV3, EfficientNetB0, EfficientNetB7, and InceptionResNetV2. To aid in capturing very minute attention dynamics, we modify these architectures and then incorporate more layers. Later, we used a model ensemble to increase the accuracy and efficiency of the model. Later, the XGBoost model is integrated with all other models used before in the hybrid model technique to obtain better accuracy and efficiency of the model. The model performance is adequately evaluated using various evaluation measures like accuracy, precision, recall, F1 Score, and support. These metrics provide a holistic understanding of the model's capability to detect and predict attention patterns in different contexts. After using the models, we could get the best accuracy from VGG19 and InceptionResNetV2, i.e., 84.6% and 83.6% respectively. VGG16 hybrid models recorded 82% in the accuracy test. With deep learning and pre-trained architectures, the Hybrid Eye Gaze Model shows a strong and flexible attention monitoring solution for varying types of applications.

Drivers' use of different types of multimedia devices while driving - a survey study

Accidents and collisions continue to be an ongoing problem in road safety. While driving, drivers are influenced by stimuli that can affect their behaviour in various ways. Nowadays, various types of multimedia devices have become popular, which, when used by drivers while driving, can influence driver attention. The literature undertakes studies of the impact of such devices on the driver's degree of distraction as measured, for example, by the driver's reaction time. A survey was conducted to identify sources of driver distraction. A total of 237 drivers aged between 18 and 78 years participated in the sample. Of the respondents, 64.6% use a mobile phone while driving, 20.3% use a smartwatch, 33.3% use a multimedia control system for vehicle functions. Respondents indicated that they had either witnessed (18.1%) or been involved (6.3%) in a collision/accident when the driver used various types of multimedia devices while driving. Of those surveyed, 2.1% had caused a collision or accident by using a multimedia device while driving. Respondents are aware of the risks of using different types of media devices while driving, but they still use them. The analyses of the survey data will be used in future research using a car simulator.

Assessing the determinants of crash propensity using structural equation modeling: Role of distractions caused by fellow drivers

Introduction: Aggressive behavior of drivers is a source of crashes and high injury severity. Aggressive drivers are part of the driving environment, however, excessive aggressive driving by fellow drivers may take the attention of the recipient drivers away from the road resulting in distracted driving. Such external distractions caused by the aggressive and discourteous behavior of other road users have received limited attention. These distractions caused by fellow drivers (DFDs) may agitate recipient drivers and ultimately increase crash propensity. Aggressive driving behaviors are quite common in South Asia and, thus, it is necessary to determine their contribution to distractions and crash propensity. Method: Our study aimed to evaluate the effects of DFDs using primary data collected through a survey conducted in Lahore, Pakistan. A total of 801 complete responses were obtained. Various hypotheses were defined to explore the associations between the latent factors such as DFDs, anxiety/stress (AS), anxiety-based performance deficits (APD), hostile behavior (HB), acceptability of vehicle-related distractions (AVRD), and crash propensity (CP). Structural Equation Modeling (SEM) was employed as a multivariate statistical technique to test these hypotheses. Results: The results supported the hypothesis that DFDs lead to AS among recipient drivers. DFDs and AS were further found to have positive associations with APDs. Whereas, there was a significant negative association between DFD, AS, and AVRD. As hypothesized, DFD and AS had positive associations with CP, indicating that distractions caused by aggressive behaviors leads to stress and consequently enhances crash propensity. Practical applications: The results of this study provide a statistically sound foundation for further exploration of the distractions caused by the aggressive behaviors of fellow drivers. Further, the results of this study can be utilized by the relevant authorities to alter aggressive driving behaviors and reduce DFDs.

Development and validation of the Driver Attention Regulation Scale: A measure of the perceived ability to regulation attention on the road

The ability to regulate attention while driving is a vital skill for remaining safe on the road. However, there is no current self-report method that measures such a construct, potentially limiting research in the area of attention and distraction. This study aimed to develop and validate a new measure, the Driver Attention Regulation Scale (DARS), that can be used to identify how drivers perceive their ability to regulate attention while driving. To achieve this aim, two online surveys were shared with unique samples of Australian drivers. Following the development of an initial item pool, principal components analysis and follow up confirmatory factor analysis were used to test the structure of the items and revealed that three factors emerged: attentional presence (e.g., focus; vigilance), attentional flexibility (e.g., awareness; shifting), and attentional capacity (e.g., processing speed; cognitive resources). Each of the subscales (α = 0.75–0.88) and the total scale (α = 0.87–0.91) exhibited a good internal consistency across the samples. Evidence of external validity was also demonstrated, as the DARS was positively associated with generalised attention regulation measures, and driver confidence, but negatively associated with emotion regulation difficulties, driver stress, and driving errors/lapses. Mediation analyses also indicated that the attentional constructs measured by the DARS may indirectly influence driving behaviours, via their effect on driver stress and confidence. Overall, the DARS has demonstrated promise as a new measure of driving-related attentional skills that can be used in future research to better understand how situational factors may impact drivers’ confidence, and also how confidence may influence driving behaviours (e.g., speeding).

UHead

The focus of Advanced driver-assistance systems (ADAS) is extending from the vehicle and road conditions to the driver because the driver's attention is critical to driving safety. Although existing sensor and camera based methods can monitor driver attention, they rely on specialised hardware and environmental conditions. In this paper, we aim to develop an effective and easy-to-use driver attention monitoring system based on UWB radar. We exploit the strong association between head motions and driver attention and propose UHead that infers driver attention by monitoring the direction and angle of the driver's head rotation. The core idea is to extract rotational time-frequency representation from reflected signals and to estimate head rotation angles from complex head reflections. To eliminate the dynamic noise generated by other body parts, UHead leverages the large magnitude and high velocity of head rotation to extract head motion information from the dynamically coupled information. UHead uses a bilinear joint time-frequency representation to avoid the loss of time and frequency resolution caused by windowing of traditional methods. We also design a head structure-based rotation angle estimation algorithm to accurately estimate the rotation angle from the time-varying rotation information of multiple reflection points in the head. Experimental results show that we achieve 12.96° median error of 3D head rotation angle estimation in real vehicle scenes.

Domain adaptation for driver's gaze mapping for different drivers and new environments

Distracted driving is a leading cause of traffic accidents, and often arises from a lack of visual attention on the road. To enhance road safety, monitoring a driver's visual attention is crucial. Appearance-based gaze estimation using deep learning and Convolutional Neural Networks (CNN) has shown promising results, but it faces challenges when applied to different drivers and environments. In this paper, we propose a domain adaptation-based solution for gaze mapping, which aims to accurately estimate a driver's gaze in diverse drivers and new environments. Our method consists of three steps: pre-processing, facial feature extraction, and gaze region classification. We explore two strategies for input feature extraction, one utilizing the full appearance of the driver and environment and the other focusing on the driver's face. Through unsupervised domain adaptation, we align the feature distributions of the source and target domains using a conditional Generative Adversarial Network (GAN). We conduct experiments on the Driver Gaze Mapping (DGM) dataset and the Columbia Cave-DB dataset to evaluate the performance of our method. The results demonstrate that our proposed method reduces the gaze mapping error, achieves better performance on different drivers and camera positions, and outperforms existing methods. We achieved an average Strictly Correct Estimation Rate (SCER) accuracy of 81.38% and 93.53% and Loosely Correct Estimation Rate (LCER) accuracy of 96.69% and 98.9% for the two strategies, respectively, indicating the effectiveness of our approach in adapting to different domains and camera positions. Our study contributes to the advancement of gaze mapping techniques and provides insights for improving driver safety in various driving scenarios.

Review on Computer Vision Techniques to enhance Road Lane Detection by using Open CV

In response to the alarming increase in road accidents attributed to driver inattention, recent advancements in road safety technology have become imperative. Addressing this concern, Lane Detection Systems have emerged as a pivotal component in technologically advanced transportation systems. These systems function by identifying lane boundaries on the road and promptly alerting the driver if there is deviation from correct lane markings. However, achieving robust lane detection poses challenges due to diverse road conditions encountered during driving, including night time and day time scenarios. This research focuses on a Lane Detection System that employs a front-facing camera to capture the road view and identify lane boundaries. The proposed method involves dividing the video image. This research presents the study explores various techniques for detecting lane markings, series of sub-images, extracting image features for each segment, and utilizing these features to recognize lanes on the road, acknowledging the complexity introduced by different driving conditions.

A review of driver gaze estimation and application in gaze behavior understanding

Driver gaze plays a key role in different gaze-based applications, such as driver attentiveness detection, visual distraction detection, gaze behavior understanding, and building driver assistance system. The main objective of this study is to perform a comprehensive summary of driver gaze fundamentals, methods to estimate driver gaze using machine learning (ML) based technique, and its applications in real world driving scenarios. We first discuss the fundamentals related to driver gaze, involving head-mounted and remote setup based gaze estimation and the terminologies used for driver gaze behavior understanding. Next, we list out the existing benchmark driver gaze datasets, highlighting the collection methodology, and the equipment used for such data collection. This is followed by a discussion of the algorithms used for driver gaze estimation, which primarily involves traditional machine learning and deep learning (DL) based techniques. The estimated driver gaze is then used for understanding gaze behavior while maneuvering through intersections, on-ramps, off-ramps, lane changing, determining the effect of roadside advertising structures and also for developing driver gaze based applications such as maneuver prediction, driver inattention, and distraction detection systems, etc. Finally, we discuss the limitations in the existing literature, challenges, and future scope in driver gaze estimation and gaze-based applications.

A Novel EEG-Based Assessment of Distraction in Simulated Driving under Different Road and Traffic Conditions.

The drivers' distraction plays a crucial role in road safety as it is one of the main impacting causes of road accidents. The phenomenon of distraction encompasses both psychological and environmental factors and, therefore, addressing the complex interplay contributing to human distraction in automotive is crucial for developing technologies and interventions for improving road safety. In scientific literature, different works were proposed for the distraction characterization in automotive, but there is still the lack of a univocal measure to assess the degree of distraction, nor a gold-standard tool that allows to "detect" eventual events, road traffic, and additional driving tasks that might contribute to the drivers' distraction. Therefore, the present study aimed at developing an EEG-based "Distraction index" obtained by the combination of the driver's mental workload and attention neurometrics and investigating and validating its reliability by analyzing together subjective and behavioral measures. A total of 25 licensed drivers were involved in this study, where they had to drive in two different scenarios, i.e., City and Highway, while different secondary tasks were alternatively proposed in addition to the main one to modulate the driver's attentional demand. The statistical analysis demonstrated the reliability of the proposed EEG-based distraction index in identifying the drivers' distraction when driving along different roads and traffic conditions (all p < 0.001). More importantly, the proposed index was demonstrated to be reliable in identifying which are the most impacting additional driving tasks on the drivers' distraction (all p < 0.01).

Effects of context-sensitive distraction warnings on drivers’ smartphone use and acceptance: A long-term naturalistic field study

Driver distraction by smartphone use is a major contributor in traffic accidents. Context-sensitive driver (in)attention monitoring and warning systems might mitigate the associated risks. However, few naturalistic studies are yet available on the effects of such systems. In this paper, utility of context-sensitivity in inattention monitoring was studied by a smartphone-based context-sensitive distraction warning system. The warnings were based on driver's phone use and the attentional demands of the upcoming traffic environment. The system's effects on 26 heavy smartphone users’ phone usage and acceptance were analyzed after a within-subject naturalistic study with 12 weeks of control (warnings off) and 12 weeks of interventions (warnings on). The system decreased odds that the drivers would touch their smartphones in reminder areas that were defined a priori as high demanding for attention. Against expectations, the system had no effect in urban road environments. The drivers reported that they had paid more attention to traffic because of the system and that the warnings were acceptable and useful, even if annoying. Similar systems’ safety effects should be further studied. No eye-tracking or driving performance measures were collected and thereby it is questionable if there was a true positive effect on participants’ attention. However, the findings suggest that (in)attention warning systems might benefit from adaptation of the warnings to the upcoming driving demands. The findings can be utilized for the development of proactive and context-sensitive (in)attention monitoring and distraction mitigation systems.

An efficient multi-task learning CNN for driver attention monitoring

Driver Monitoring System (DMS), usually equipped with a camera, is an emerging vehicle safety system that can monitor driver attentiveness and trigger timely alarms when signs of inattention are detected. Since a single indicator (e.g., eye blink rate) is insufficient and unreliable to analyze driver attentiveness, almost all existing solutions train several independent models to identify driver facial states, such as face landmark, head pose, yawning, eye state, etc. However, apart from neglecting the inherent correlations between these related tasks, multiple models also raise challenges for vehicle safety-critical systems (e.g., hardware resources, software compatibility, and real-time response). In this paper, we propose a multi-task learning CNN framework (DANet) to unify the relevant tasks into one model and simultaneously output various driver facial states. By sharing the common features and parameters of highly related tasks, DANet avoids repetitive computations and mitigates single task overfitting. More importantly, the model provides a comprehensive overview of facial states while maintaining low complexity. We also propose two novel designs: (1) Dual-loss Block, which decomposes the pose estimation task into pose classification and coarse-to-fine regression; (2) Head Pose Penalization, which constrains the network to predict gaze direction based on predicted head pose. Our method achieves compelling results in both speed and accuracy on a vehicle computing platform, marking a momentous step in this field.