Detection of Safety Features of Drivers Based on Image Processing

Xi, Haoning; Zhang, Yi

doi:10.1061/9780784481523.208

Cited by 8 publications

(2 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In [50], the authors present a driving safety features scheme (referred to as DSF in Figure 6) for identifying abnormal driving features to enhance safety. DSF employs the histogram algorithm with an SVM classifier.…”

Section: A: Mathematical Models-based Schemesmentioning

confidence: 99%

AI-Driven Driver Behavior Assessment Through Vehicle and Health Monitoring for Safe Driving—A Survey

Yaqoob,

Morabito,

Cafiso

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Safe driving is a serious and challenging concern that is dependent on driver behaviors including aggressive, distracted, fatigued, and drowsy driving. In a smart society, in-vehicle monitoring of drivers and detecting abnormal driving behavior as anomalies can reduce the rate of road crashes. Existing surveys cover the various schemes for detecting on-road driver behaviors through sensing data. We have identified a gap in the linkage of driver's health or vehicle conditions to abnormal behaviors that are not yet covered. This work presents a taxonomy of schemes, and the analytical evaluation and identifies the open research challenges. The work specifically investigates the modeling of driver behavior and the detection of abnormal behavior, utilizing techniques such as AI-based image processing, signal processing, and traditional algorithmic approaches. This analysis encompasses methods and features based on both driver health and vehicle monitoring with the ultimate goal of ensuring safe driving. More specifically, existing approaches are classified in a coherent taxonomy by reviewing traditional mathematical, machine learning, and deep learning-based schemes. Moreover, a summary of schemes is presented to highlight the key points followed by a comprehensive analytical discussion. It aids in pinpointing research issues by leveraging the insights gained from the comparison. Finally, the work highlights the open research challenges for the researchers to provide innovative solutions.

show abstract

Section: A: Mathematical Models-based Schemesmentioning

confidence: 99%

AI-Driven Driver Behavior Assessment Through Vehicle and Health Monitoring for Safe Driving—A Survey

Yaqoob,

Morabito,

Cafiso

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Many pretext tasks were found to be conducive to learn image features. For example, image colorization [40,41], super-resolution [21], image processing [39,3], jigsaw puzzles [26], rotation angle prediction [12] and unsupervised deep clustering [2,34]. These methods can learn desired and transferable representations that achieve promising results in downstream tasks.…”

Section: Self-supervised Learning With Pretext Tasksmentioning

confidence: 99%

Contrastive Learning with Diverse Samples

Wu,

Liu

2023

Frontiers in Artificial Intelligence and Applications

View full text Add to dashboard Cite

Unsupervised visual representation learning has gained much attention from the computer vision community because of the recent contrastive learning achievements. Current work mainly adopts instance discrimination as the pretext task, which treats every single instance as a different class (negative) and uses a collection of data augmentation techniques to generate more examples (positive) for each class. The idea is straightforward and efficient but will generally cause similar instances to be classified into different classes. Such problem has been defined as “class collision” in some previous works and is shown to hurt the representation ability. Motivated by this observation, we present a solution to address this issue by filtering similar negative examples from each mini-batch. Concretely, we model the problem as a Determinantal Point Process (DPP) so that similar instances can be filtered stochastically, and diverse samples are expected to be sampled for contrastive training. Besides, we further introduce a priority term for each instance, which indicates the hardness of its positives, so that instances with more hard positives are more likely to be sampled for contributing to the optimization. Our sampling can be efficiently implemented in a feed-forward manner and further accelerated by our encouraged complement DPP. Extensive experimental results demonstrate our priority over the standard setup of contrastive learning.

show abstract