Fig 2 - uploaded by Daniela Ndunguru
Content may be subject to copyright.
![Real driving experiment [28]](profile/Daniela-Ndunguru-2/publication/379585326/figure/fig1/AS:11431281303406471@1736842381134/Real-driving-experiment-28.png)
Source publication
![Fig.2 Real driving experiment [28]](profile/Daniela-Ndunguru-2/publication/379585326/figure/fig1/AS:11431281303406471@1736842381134/Real-driving-experiment-28_Q320.jpg)
Understanding and predicting drivers' gaze patterns is essential for improving road safety and optimizing in-vehicle displays. This study delves into the nuanced dynamics of drivers’ visual attention across varied road segments, employing both statistical analyses and machine learning models. Ten participants, spanning diverse demographics, partici...
Context in source publication
Context 1
... was conducted at the Weishui Campus of Chang'an University in Xi'an, China, specifically designed to meticulously collect data on drivers' gaze behavior. The chosen vehicle for this experiment was a standard Volkswagen Sagitar, featuring an automatic transmission, a 1.6-L engine, a 3-box design, and seating for five passengers, as depicted in Fig. 2. This vehicle choice aimed to simulate a typical private vehicle, aligning with real-world driving scenarios. Participants, totaling ten in number, actively engaged in the experiments, forming a diverse group of four females and six males. The age range spanned from 21 to 60 years, with a mean age of 35.53 and a standard deviation of ...
Similar publications
This paper presents the myEye2Wheeler dataset, a unique resource of real-world gaze behaviour of two-wheeler drivers navigating complex Indian traffic. Most datasets are from four-wheeler drivers on well-planned roads and homogeneous traffic. Our dataset offers a critical lens into the unique visual attention patterns and insights into the decision...
This paper presents the myEye2Wheeler dataset, a unique resource of real-world gaze behaviour of two-wheeler drivers navigating complex Indian traffic. Most datasets are from four-wheeler drivers on well-planned roads and homogeneous traffic. Our dataset offers a critical lens into the unique visual attention patterns and insights into the decision...
Citations
Road accidents are one of the major contributors to mortality in India with nearly 1.68 lakh deaths in 2023 accounting for nearly 19 deaths per hour. While road accidents aren’t always “fatal”, delayed detection leads to the loss of many lives. Deep Learning deliberately showed terrific results for a lot of tasks, but for a long time, the Deep Learning models utilized the standard Multi-Layered Perceptron (MLP) architecture as its’ skeleton. Recently, a state-of-the-art architecture—Kolmogorov Arnold Network (KAN) showed up, with variable activation functions in each edge of the graphical representation of the network. Several works have inherited the KAN architecture and have shown better results in many instances. This research proposes a class of accident detection models, SADAK (Simple, and Automatic Detection of Accidents on roads using Kolmogorov–Arnold Networks) for detection of road accidents inheriting KAN architecture. To benchmark its efficiency, several well-known MLP based architectures have been considered, alongside some state-of-the-art models, which result in accuracies as high as 89% using the MLP (Vision Transformer), and 97% using the KAN architecture (KAN-Convolutional-MLP).
Accurate classification of eye state using Electroencephalogram (EEG) data reflects the high dimensionality and intricacy associated with the nature of the EEG signals. This state-of-the-art technique applies Discrete Cosine Transformation (DCT) for feature extraction, Cuckoo Search Algorithm (CSA) for optimal feature selection followed by a robust stacking algorithm integrating Adaptive Gradient Boosting Classifier (A-GBC) and k-Nearest Neighbors (k-NN) with Logistic Regression as meta-learner to improve the predictive accuracy and model performance. This model would strengthen the accuracy, reliability, and efficiency of eye state classification. DCT is used to compress the time series data into frequency domain for feature extraction, resulting in concentrating a vast amount of superficial signal energy. The next stage uses the Cuckoo Search Algorithm (CSA) in the selection of the optimal features. It thus optimizes the feature sub- set of the input in-order to get an optimal learning rate. The combination of these processes with a stacked ensemble using Adaptive Gradient Boosting Classifier (A-GBC) & k-Nearest Neighbors (k-NN) basal learners results in an immediate classification following the path of a Logistic Regression meta learner that achieves a remarkable classification accuracy of 96.96%. This strongly emphasizes the exceptional efficiency of the proposed framework when placed in context with existing approaches. It implies that the research not only reinforces the existing state of EEG eye state categorization but also provides for the optimization of its utility in other types of biomedical signal processing, thereby making it an example of a versatile medical tool.
