Nighttime driving presents significant challenges, including detecting pedestrians, cyclists, and potholes in low-visibility conditions, leading to frequent road accidents. In Sri Lanka, there is a lack of automated systems capable of addressing these challenges. This research proposes a system leveraging AI and real-time video processing to enhance driver safety by improving pedestrian and cyclist visibility and developing a reliable pothole detection system for night time. These enhancements rely on extensive data collection and preprocessing for training specialized computer vision and machine learning models tailored for real-time, low-light scenarios. The application provides real-time alerts and visual cues to reduce nighttime driving accidents. Additionally, a software framework is proposed to offer real-time feedback for drivers leveraging available hardware capabilities, enhancing awareness of vulnerable road users. Through rigorous data collection, testing, and development, this approach significantly helps to enhance nighttime driving safety in Sri Lanka.

The escalating prevalence of work-related stress has led to a notable decline in work performance and the mental well-being of the workforce. Studies suggest that personally tailored music can have a positive impact on reducing stress levels. This paper proposes a novel approach to generate personalized music by combining brain activation analysis and musical preference analysis. The proposed method involves three main stages: (1) Source Separation, (2) Brain Activation Analysis, and (3) Personalized Music Generation. In the first stage, we use a variant of the Wave-U-Net to decompose the input song into its vocals and melodies. Subsequently, we employ electroencephalography (EEG) to analyze the user’s brain activation and identify preferred segments. The third stage involves the generation of personalized music using a transformer-based music generation model, which takes into account the user’s music preference and brain activation patterns. Furthermore, this approach was assessed through an experiment with 10 participants, achieving an average satisfaction level of 3.9/5 and a 52.97% increase in Frontal Alpha Asymmetry (FAA) fluctuation percentage after analyzing brain activation with the generated melody. As future work, we intend to conduct intervention-based experiments with large sample sizes and expand into evaluation metrics that provide objective measures of work stress reduction.

Cardiopulmonary Resuscitation (CPR) is an essential skill for medical professionals. Currently, CPR training is carried out using mechanical manikins which have drawbacks when it comes to the realism of the simulation, cost and durability. In the meantime, Virtual Reality (VR) has entered a new phase of wider availability and affordability in the consumer market. Therefore, a low-cost VR-based solution can now be proposed to address some of these issues with the mechanical manikin. Hence, this study presents an approach where the mechanical manikin has been augmented with VR. To test the viability and acceptance of this solution, a user based evaluation was carried out with a group of experts and novices in CPR. Despite their contrasting experience and knowledge of CPR, both groups have expressed favourable views towards the basic solution presented in this paper.

Typeface spacing is a hard problem. It takes countless hours of manual labour to achieve an aesthetically pleasing font one frequently encounters in digital media. Inter-letter spacing defines the texture and the feel of a typeface and when done accurately yields an aesthetically balanced and an appealing typeface. Nevertheless, setting spacing in a typeface is a tedious and a time consuming task. Hence this paper presents an exploratory study investigating the potential of Neural Networks (NN) to fully automate the typeface spacing process. Even though the NN models investigated in this study yielded up to an accuracy of 47\% when compared with typefaces spaced by the type designers, the visual differences were subtle. Thus, we conclude that neural models can indeed be used to model the typeface spacing problem. As one of the first attempts to apply neural models in this particular problem domain, this study lays the foundation to future research and studies.