University of Wah Journal of Computer Science

A Dual Framework of Feature Selection based on the Fusion of HOG and Pyramid HOG for the Categorization of COVID-19

2022-08-30T04:20:09+00:00

COVID-19 was initially detected in Wuhan, China. The virus spread all over the world at a rapid speed. COVID-19 is an infection that may cause infections in the respiratory system and the lungs. In order to diagnose COVID-19, chest X-rays have been utilized extensively. The purpose of this research is to create a computer-vision-based method for identifying COVID in chest X-rays. In the proposed model handcrafted features known as HOG and PHOG are derived and fused. After features have been fused, the Binary Grey Wolf Optimization technique is used with an Entropy-Based Optimization Algorithm to select the most significant features possible. The proposed model results are evaluated on a benchmark X-ray dataset that gives greater than 99% accuracy. The proposed model performed better compared to existing published works in this domain.

Classification and Detection of Pathogens from Enhanced Microscopic Images

2022-09-09T02:54:15+00:00

Drinking water is essential for human life but unfortunately every year, many lives are lost as a result of the use of polluted water. Computerized methods play a dynamic role in detecting pathogens from water. The first symptoms of pathogens in water are difficult to detect by the naked eye at an early stage. Therefore, in this research, a computerized method is proposed in which features are extracted from the pre-trained ResNet-50 model, and classification of the different types of pathogens is performed using the SoftMax layer. The proposed method's performance is evaluated on the proposed microscopic pathogen dataset. The proposed dataset is pre-processed using Enhanced Super-Resolution Generative Adversarial Networks (ESRGAN) method for the enhancement of Image quality. The proposed method provides greater than 90% prediction accuracy.

An Intelligent Health Control Security Robotic System

2022-10-10T06:14:55+00:00

Nowadays, scientific knowledge is constantly bringing comfort and change to everyday life but the entire world is facing a major health crisis as a result of coronavirus disease transmission. According to the World Health Organization (WHO), wearing a mask on the face in public areas is one effective method of protection against COVID. Autonomous robots have become a prominent technology in recent years, with applications in a variety of fields. Robots are utilized to complete tasks more quickly than humans. Generally, robots are smarter with endless energy levels, and are precise in task management. Therefore, propose the methodology of this work is focused on developing a robot-based COVID-19 protection system. The proposed robot-based COVID-19 protection system consists of five core steps: 1) Person Identification, 2) Vaccination Checking 3) Face Recognition 4) Face Mask Detection, and 5) Temperature Checking. Person identification is performed using HOG to detect faces and machine learning classifier SVM for identification of the person. Then vaccination status is checked. To check vaccination status, it gets the name from face recognition and matches with the names in the database and gets a value from the vaccination column to show if the person is vaccinated or not and facemask detection is performed by facemask detector using Keras Mobile-Net architecture. The temperature is checked using the MLX90614 temperature sensor. The robot performs all these functions by speaking and by displaying them on the LCD screen. Artificial Intelligence on basis of deep learning and neural networks could help in fighting Corona Virus in many ways. The purpose of this system is to use a security robot instead of a security guard in organizations such as universities, colleges, schools, offices, software houses, and other organizations. The proposed system performed better compared to the existing systems as it achieves 99% precision and a 0.01% error rate.

A Systematic Review on Ovarian Cancer

2022-12-16T04:35:26+00:00

Most women suffer late-stage ovarian disease endure a high pace of mortality. It is required to identify and analyze cancer right on time in its initial phase of development. It is generally suggested clinical screening of women through different detection techniques like imaging modalities (ultrasound, CT scan, MRI) and also detection using different biomarkers for cancer symptoms existing in patient’s blood. Biomarker, a distinguished device that are available for patients with ovarian malignant growth, the particular one is cancer anigen CA-125, is usually tried for medical use which is supposed to be more efficient biomarker to detect the infection. Here, we depict elective biomarkers like CA-125 which conquer a significant number of the issues related to malignancy, for example, expanded affectability and specificity, particularly in the beginning phases of ovarian cancer, and which could be utilized effectively in a biosensor design.

An Advance Driving Assistant System Simulator using Unity-3D

2022-08-30T05:05:58+00:00

Simulations are 3D environments that are based on real-world scenarios. As technology is evolving in automobiles, the need for advancement in Artificial Intelligence (AI) of automobiles and to train AI systems of the automobiles are getting expansive. This study investigates the problem of collecting reference data for the testing and evaluation algorithms for autonomous vehicles using unity simulation. Using computer simulations is one way to solve this issue. A genuine system modeled using computer simulations, including all of its static and dynamic properties. There are 4 modules: forward & backward collision, speed warning, traffic light & sign detection, and lane departure system. All the modules are implemented in C# language and the unity-3d platform. The composite collider module is used for the lane departure in which the line render algorithm is applied for forwarding/backward collision. This method reduces the testing period while ensuring accuracy and efficiency in data collection. The construction of a simulation environment based on Unity with the ability to test sensors and algorithms for autonomous cars and display deviations from reference data is the goal of this research work. Roads, sidewalks, buildings, traffic signs, and automobiles are among the common city players, and items in the simulation model are presented in the simulation. The simulation is tested by the methods like black box testing, use case testing, performance testing, stress testing, module testing, Software-In-the-loop, Driver-in-the loop testing.