Theses - Undergraduate

Browse

Recent Submissions

Now showing 1 - 5 of 325
  • Item
    Open Access
    Prediction of Anxiety Level Based on Electrocardiogram (Ecg) Images in the Context of Bangladesh
    (North South University, 2023) Tahrima Ihsan; Asfaria Islam Chowdhury; Syed Fardeen Mehrab; Eftekhar Masud; Nova Ahmed; 1831819642; 1931741042; 2031757642; 1812459042
    Heart health conditions and mental health status have been proven to be linked in many cases. In the past decade, there has been an increase in heart health diseases and declining mental health reports among the youth of Bangladesh. Due to the high rate of illiteracy or unawareness in mental health conditions, many people who suffer from anxiety or anxiety related disorders do not seek professional help which may or may not lead to serious heart health diseases later in life. A better understanding of an individual’s heart and mental health status can prove to prevent a young adult from suffering from a life-threatening condition in their later years. This assessment can be done through analyzing electrocardiogram (ECG) scans and its correlation with the individual’s anxiety scale score evaluated from a standardized questionnaire which has been adapted in the context of Bangladesh. Using deep neural network models, the percentage of correlation can be unfolded.
  • Item
    Open Access
    Violence Detection Using MobileNetV2
    (North South University, 2023) Shazia Mahazaben; Iftekhar Md Ruddro Hasan; Sayma Islam; Abu Obaidah; 1912916642; 1913010642; 1912673042
    This project aims to develop a violence detection system using the MobileNetV2 deep learning architecture. Violence detection has gained significant attention in recent years due to the increasing need for automated surveillance systems in public spaces and security applications. The MobileNetV2 architecture, known for its efficiency and accuracy in image classification tasks, is leveraged to build a robust and real-time violence detection model. The proposed system consists of two main stages: training and inference. In the training stage, a large dataset of annotated images and videos containing violent and non-violent scenes is used to fine-tune the MobileNetV2 model. Transfer learning techniques are employed to adapt the pre trained network's weights to the violence detection task, enabling the model to learn discriminative features related to violent behavior. During the inference stage, the trained model is applied to real-time video streams or pre recorded videos. The MobileNetV2 architecture, with its lightweight design, allows for efficient processing on resource-constrained devices such as smartphones and embedded systems. The model analyzes each frame or a sequence of frames, extracting relevant features and making predictions on the presence of violence. The performance of the violence detection system is evaluated using standard metrics such as accuracy, precision, recall, and F1-score. Extensive experiments are conducted on benchmark datasets and real-world scenarios to assess the system's effectiveness and reliability. The results demonstrate the capability of the MobileNetV2-based violence detection system to accurately identify violent incidents while maintaining real-time performance.
  • Item
    Open Access
    Digitation of an Analog Kymograph Machine
    (North South University, 2023) Sadman Uddin Ahmed; All Mehedi Saqlain; Mahmud Hasan Mredul; Atiqur Rahman; 1831518043; 1912120643; 1912926643
    Biomedical research has been altered by the digitization of experimental setups, such as the organ bath and kymograph, by overcoming the drawbacks of conventional approaches. Potential inaccuracies and limitations in data collecting and analysis have been caused by manual measurements and subjective interpretations. The goal of this project is to create a dependable, user-friendly digital system that combines an accelerometer, an organ bath, and a kymograph. Real-time data capture, automatic recording, and enhanced visualization capabilities can all be accomplished by observing and analyzing mechanical movements. Accelerometer integration automates data collection, improving accuracy and productivity. Digitization has made it possible to use sophisticated visualization techniques that provide accurate and thorough depictions of physiological changes. Algorithms and software technologies allow for deeper insights into underlying processes. The precision, effectiveness, and dependability of physiological research have all significantly improved with the integration of accelerometers into experimental setups. For continued development and use of digitized systems, however, issues with accelerometer measurements, setup integration, and protocol standardization must be resolved.
  • Item
    Open Access
    Optical Characterization and Propulsion of Macroscopic Objects Using Nanostructures for Self-stabilization
    (North South University, 2023) Puja Saha; Raiyan Mahmuda; Mahdy Rahman Chowdhury; 1731075043; 1831037643
    Our project is about an unfocused beam of light which can provide enough intensity to balance the weight of the nanoscopic reflecting structure. We are proposing an elementary building block consisting of unit cells made by silicon (n=3.5) as optical resonators of sub wavelength thickness on a thin substrate silica (n=1.45). The unit cells are in a mirror-symmetric manner with respect to the structure’s center of mass, to eliminate the moment and lateral force. The asymmetry in the optical response can restore torque which causes the displacement from its equilibrium position or orientation. We have got our result in COMSOL Multiphysics Software where we have proposed MEPS design in a mirror symmetric manner consisting of two dissimilar rectangular resonators.
  • Item
    Open Access
    AI based patient monitoring robot
    (North South University, 2021) Debnath Banik; Mohammad Ashrafuzzaman Khan; 1410118043
    Bangladesh as a developing country still is facing the limitation of health care facilities. Large population and outbreak of diseases, and other factors made the medical system struggling and tougher to full fill the meets and for that many patients are not getting the proper health care for less hospitals and beds and less medical stuff per ratio to population. So, introducing a robot in medical sector may bring a helping hand to the nurse doctors of hospitals. Specially in the general word of the hospitals where very less number facilities of advance health equipment and machine is also a big problem. The outbreak of any disease also brings a big challenge to the medical sector. The number of patients is so large that hospital can’t arrange available seats for them getting no option the doctor give treatment all the extra patient arranging bed in the Flore. Monitoring and serving the patient get tougher for medical stuffs with the increase number of patients. Here if we include a health monitoring and assisting robot which may take health update with time to time can bring a temporary solution into crisis in number of expensive health equipment which are basically stationary and heavy to move around. Making the long communication within doctor and junior doctors or nurse with the robot may make treatment more effective and other advanced facilities of the robot may come handy for the hospital management stuff to perform the duty more efficient and easily.