Dunhill Guantero

Addressing the challenge of building sustainable, clean communities has become increasingly urgent in the face of rising waste and pollution due to growing populations. This surge in rubbish not only escalates pollution levels but also negatively impacts our neighborhoods. To combat this issue, our study introduces an innovative system designed to detect plastic waste in rivers using the YOLOv5 algorithm. We utilized a Raspberry Pi Model 4 B as the core microcontroller, paired with a 5MP Camera Module and a USB camera to capture images of floating plastic bottles.

The algorithm was trained on a custom dataset, meticulously processed on a computer. The results are promising: our model achieved an impressive 84.298% accuracy in identifying plastic bottles in river settings. Additionally, with a precision rate of 79.14% and a recall rate of 57.37%, our system demonstrates significant effectiveness in object detection, marking a substantial step towards cleaner, waste-free waterways.

In commercial establishments, maintaining controlled restroom access is essential for both security and customer satisfaction. Unauthorized entry can lead to loitering and other issues, detracting from the establishment's overall environment. To address this challenge, I developed a sophisticated barcode-based door lock system using the ATmega328P microcontroller. This innovative solution integrates key components such as a barcode scanner, servo motor, and LED indicators, all working together to create a seamless and secure access mechanism tailored to the needs of commercial environments.

The system's controls are designed to be both straightforward and highly efficient. Authorized users simply scan a valid, non-expired barcode to gain access, with the system processing the input swiftly and unlocking the door when appropriate. Additionally, a tactile button provides a manual override option, ensuring flexibility in access. Throughout the development process, careful consideration was given to the system's usability, with minor response delays being minimized to maintain a smooth user experience.

By restricting restroom access to authorized individuals, this system offers a robust solution that significantly enhances restroom security and operational efficiency. It not only deters unauthorized access but also contributes to a cleaner, more pleasant environment, thereby boosting overall customer satisfaction.

In developing this system, I took a hands-on approach, coding in C# for the ATmega328P microcontroller and meticulously integrating the barcode scanner, servo motor, and other critical components. The circuit design, crafted from the ground up using EasyEDA and Proteus, was a key focus, ensuring both security and user-friendliness in the final product. The PCB was designed with precision, resulting in a reliable, easy-to-use system that prevents unauthorized entry, ultimately enhancing cleanliness and customer satisfaction within commercial establishments.

I played a pivotal role in transforming the digital experience for an international client by authoring and designing an intuitive User Guide webpage. This project was more than just a web page—it became a crucial resource that significantly improved the end user’s journey, providing them with detailed support and comprehensive resources to navigate the company’s offerings with ease. To kickstart the project, I constructed the initial version of the webpage using WordPress, leveraging the power of the Elementor plugin. This approach allowed me to create a flexible and easily manageable site structure that could adapt to the client’s evolving needs. However, the project didn’t stop there. To ensure greater customization and performance, I meticulously converted the final version into pure HTML, CSS, and JavaScript, giving the site a robust, lightweight foundation that would stand the test of time.

In addition to the User Guide, I developed a sleek and responsive landing page, along with dynamic, interactive portal pages. Using Vanilla CSS, Bootstrap, and jQuery, I crafted a visually appealing interface that not only welcomed users but also provided a seamless and engaging navigation experience. These enhancements ensured that different user groups, regardless of their technical proficiency, could effortlessly access the information and tools they needed, solidifying the client’s online presence as both user-friendly and professional.

Pneumonia and COVID-19 are among the most severe respiratory diseases, often leading to high morbidity and mortality rates worldwide. In the fight against these life-threatening conditions, early and accurate diagnosis is crucial, enabling timely medical interventions and significantly improving patient outcomes. Recognizing this need, our study embarks on developing and comparing two advanced hybrid models—Convolutional Neural Network-Gated Recurrent Unit (CNN-GRU) and Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM)—designed to detect pneumonia and COVID-19 from chest X-ray images with high precision.

By harnessing the power of these deep learning techniques, the study aims to push the boundaries of medical image analysis, offering new insights that could revolutionize the diagnostic process for respiratory infections. The CNN-GRU and CNN-LSTM models were meticulously trained and tested on a diverse dataset of chest X-rays, allowing us to assess their performance in detecting the subtle yet critical patterns associated with these diseases.

The results were compelling. A t-test comparison revealed that the CNN-GRU model outperformed the CNN-LSTM model in terms of accuracy, with a t-statistic of 9.3749—significantly higher than the t-critical value. This finding not only underscores the superior performance of CNN-GRU in this context but also highlights its potential as a more effective diagnostic tool in managing infectious diseases.

Through this research, we aim to contribute valuable knowledge to the medical community, paving the way for the development of more reliable and efficient diagnostic systems that could ultimately save lives by enabling quicker, more accurate detection of severe respiratory conditions like pneumonia and COVID-19.

I developed a unique and interactive two-player strategy game using React.js. The core of the game revolves around Players A and B, who each have six numbered pieces, ranging from 1 to 6. These pieces are placed strategically on a 3x3 grid, but there’s a twist: larger pieces can cover smaller ones, adding an exciting layer of strategic depth. The goal is to outsmart your opponent by aligning three of your pieces in a row, column, or diagonal, while staying within the constraints of the game’s mechanics. To keep the pace brisk, each player has only 15 seconds to make a move, or their turn is skipped, adding an element of time pressure that encourages quick thinking and adaptability.

In building this game, I created five core components: Board.js, which dynamically manages the game’s grid and tracks the movement of the pieces; HowToPlay.js, a user-friendly guide that explains the rules; Loader.js, which ensures smooth transitions and seamless navigation; Menu.js, which offers intuitive navigation and game controls; and PieceSelector.js, which allows players to select and place their pieces during their turn. The entire application is styled using vanilla CSS, with a focus on clean, minimalist design. I also incorporated subtle sound effects for interactions like button presses to enhance the user experience.

One of the key aspects of the project was ensuring that the game is fully responsive, providing a consistent experience across all screen sizes—from desktops to mobile devices. I paid special attention to responsive design techniques, ensuring the game’s layout adjusts seamlessly no matter the device, so players can enjoy it from any platform. This project demonstrates my ability to develop interactive, well-structured applications using React, focusing not only on functional gameplay but also on responsive design and user experience.