Journal of Advanced Research in Computing and Applications

Face Detection based on Haar Cascade and Convolution Neural Network (CNN)

Noor Afiza Mohd Ariffin — 2025-03-10

Face detection plays a crucial role in identifying individuals during suspicious activities, serving as a foundational component in various security applications. Modern face detection systems leverage machine learning algorithms to accurately identify human faces in images or videos, facilitating authentication in security contexts. This paper presents an innovative face detection system that integrates the Haar cascade method with Convolutional Neural Networks (CNNs), aimed at enhancing the accuracy of facial detection. The evaluation of the proposed system was carried out in a Python environment, utilizing real images from well-established public datasets, including Faces94, Faces95, Faces96, and the Grimace dataset, curated by Libor Spacek. The results obtained demonstrate the efficacy of the integrated approach, achieving accuracy rates of 98.37%, 97.22%, 97.52%, and 100% for the Faces94, Faces95, Faces96, and Grimace datasets, respectively. These findings indicate that the combination of Haar cascade and CNN-based methodologies significantly outperforms traditional machine learning face detection techniques, underscoring the potential for improved accuracy in real-world face detection applications. This research contributes to the ongoing advancements in facial recognition technology, with implications for enhanced security measures and intelligent human-computer interaction.

Analysis of Different Deep Learning Networks for Estimating the Tensile Strength of Self-Compacting Concrete Containing Recycled Aggregates

Jesús de Prado Gil Prado Gil — 2025-03-10

This writing evaluates and contrasts the performance of various deep learning models in predicting the tensile strength of self-compacting concrete incorporating recycled aggregates. Experimental data sourced from existing literature were used to create test, training, and validation sets. A range of artificial intelligence models and optimization algorithms were explored to train these networks, with adjustments made to their architectures and parameters. The results revealed patterns that offer valuable insights into the relative efficacy of the models, advancing the understanding of how deep learning can be applied to predict concrete properties. This study serves as a strong reference point for both researchers and professionals in the building industry.

Theoretical Analysis of Two-Phase Mixed Convection Flow: Effects of Fluid-particle Interaction and Mass Concentration on Velocity, Temperature and Skin Friction

Nur Hazwani Zamri — 2025-03-10

Research on Newtonian fluids, a fundamental area in fluid flow studies, serves as a key reference for more complex models. The deformation of fluid structure due to the presence of solid particles in Newtonian fluids necessitates a detailed analysis. Conventional models are insufficient to describe such systems, as they are categorized as two-phase fluid flows. The study of two-phase systems (fluid and solid) has gained significant attention due to its wide applications in real-life scenarios, including air and water pollution, blood flow in arteries, and coolant flow in vehicle engines. This field's high potential has driven researchers to find effective solutions to these challenges. The two-phase model not only focuses on fluid flow but also accounts for the interaction between the fluid and solid particles. This study theoretically examines a modified model that considers fluid phase and dusty phase under mixed convection case. Using an appropriate similarity transformation, the governing partial differential equations are converted into ordinary differential equations. The Runge-Kutta-Fehlberg method, implemented in the Maple software, is used to solve the obtained equations numerically. The study explores various physical parameters, including mixed convection parameter, fluid-particle interaction parameter and mass concentration parameter for both phases. The findings reveal that the presence of dust particles significantly affects the velocity and temperature behavior of both phases. Specifically, increasing particle concentration leads to a reduction in fluid velocity and temperature distribution.

Path Navigation in Designated Environment using Modified TOR

A'Qilah Ahmad Dahalan — 2025-03-10

Research in the field of path navigation efficiency has continuously progressed, particularly focusing on generating collision-free paths for agents moving within specific environments. Therefore, this study aims to tackle the path navigation problem using a numerical iterative approach. An improvise approach namely Modified Two-Parameter Over-Relaxation (MTOR) is introduced to address navigation challenges effectively. In numerical simulations, MTOR is utilized to derive solutions to Laplace's equation, known as harmonic functions. These harmonic functions are then used in a gradient descent scheme to generate smooth and collision-free paths for agents moving through the designated environment. The formulation of the MTOR iterative method is explained in detail, and multiple numerical experiments and simulations are carried out to assess its efficiency. The findings indicate that the proposed improvise method outperforms existing methods.

UX Evaluation of Dyslexia App Based on Honeycomb Model

Siti Asma Mohammed — 2025-03-10

Dyslexia is a learning disorder that affects literacy skills, creating challenges for children in traditional educational settings. This study evaluates the overall user experience (UX) of Mari Membaca app, which is designed to help dyslexic children improve their reading abilities. Using the UX Honeycomb model as a framework, the app was evaluated through usability testing tasks and teacher interviews. The findings show that the app supports letter recognition and reading comprehension while keeping children engaged with interactive features. However, there is room for improvement, such as using more dyslexia-friendly fonts, adding foundational modules, and expanding phonics content. This research highlights how user-centered design and structured feedback can enhance educational tools, making them more inclusive and effective. The study also emphasizes the potential of digital solutions like Mari Membaca to transform learning for dyslexic children and calls for their integration into educational policies to promote accessibility.