Summary: Developed a custom Deep Convolutional Neural Network (DCNN) model for Bengali Sign Words classification tasks, achieving a notable accuracy of 99.12% on my constructed primary dataset.
Background: This project was part of a research initiative to explore optimization techniques for computer vision tasks using DCNN-based architectures.
Skill Set: Machine learning, data preprocessing, deep learning frameworks (TensorFlow), and performance evaluation metrics.
Contribution: Sole developer of the model and optimizer; conducted comparative analysis.
Link: GitHub Repository
Summary: This system employs four well-performed transfer learning techniques named VGG16, VGG19, AlexNet and InceptionV3 with pre-trained weights. The models show the training accuracy of 99.92%, 99.58%, 98.70% and 97.86% for VGG16, VGG19, InceptionV3 and AlexNet respectively whereas validation accuracy is 92.41%, 91.62%, 88.22% and 84.95% for VGG16, VGG19, InceptionV3 and AlexNet respectively.
Background: After completing a basic research project related to COVID-19, one of my faculty advised me to work on Bengali Sign Words as there were relatively rare work on this field. I started and completed it.
Skill Set: Computer Vision, Transfer learning techniques, research writing, and data analysis.
Contribution: Lead author; designed and implemented the model, and conducted result validation. Independently written the manuscript.
Link: GitHub Repository
Summary: This is the first dataset about sign words of BdSL. This dataset is developed by collecting data from people. This is an image dataset and is a collection of 1105 images of sign words. A total of 11 sign word categories are selected which are important and daily use in our life. As this is an image dataset, so the images of sign words are taken by camera from the sign users of Bangladesh
Background: There were lots of datasets of BdSL sign alphabets, numbers, or characters, there are not enough datasets of sign words. So I initiated to collect and construct a dataset with a teacher guidance of my undergraduate institute.
Skill Set: Data visualization and analysis, Python programming, and data collection.
Contribution: Developed core analytics functions and visual components for actionable insights. Sole contribution in whole project.
Link: Mendeley Data Repository
Summary: Using the current COVID-19 pandemic data, we will analyze and predict the behavior of the COVID-19 pandemic using machine learning algorithm.
Background: This is an initial research project by me, initiated just after completing the basic of research and learning AI fundamentals during COVID-19 pandemic to practice my research article writing. I improved my writing and skills while completing this project.
Skill Set: Open-source collaboration, Python programming, and manuscript writing.
Contribution: Sole contribution by developing models to manuscript writing.
Work 5: Diseases classification using transfer learning techniques and ensemble models (Ongoing)
Summary: This is an ongoing collaboration project with university faculty on eye disease and ovarian cancer subtypes classification using transfer learning pretrained models.
Background: This project is part of a collaboration with one of my university faculty. The faculty invited me to join this project and I accepted his proposal and contributed the project.
Skill Set: AI modeling, Python programming, data collection, and data visualization.
Contribution: Designed the AI algorithm and collaborated with programmers to implement. The manuscript is written independently by me.