rabiul02
B.TECH 3RD YEAR STUDENT
About me
Data Analyst and Developer specializing in algorithm-driven solutions and efficient data processing. Skilled in Python, SQL, and core computer science concepts including data structures, dynamic programming, and optimization techniques. Experienced in implementing search and sorting algorithms, working with large datasets, and building scalable data pipelines. Proficient in data visualization tools such as Power BI and Tableau to convert complex data into clear insights. Strong problem-solving ability with a focus on performance, accuracy, and real-world application development.
Photos
Work Experience
Handwritten digit recognition using TCNN
University of engineering and management kolkata
March 16, 2024
-
November 28, 2025
HOG Feature-based Devanagari Handwritten Digit Recognition System(July 2025-Dec 2025 ) • Developed handwritten digit recognition system using HOG features with three-layer CNN achieving 99% accuracy • Implemented data preprocessing, feature extraction pipeline, and optimized CNN architecture in Python • Tools Used: Jupyter notebook, HOG
Education
Pursuing B.Tech in CS
8
2023
-
Present
B.Tech 3rd Year student with a strong foundation in Data Analysis and Software Development. Skilled in Python, SQL, and core concepts of Data Structures and Algorithms, including problem-solving, optimization, and efficient coding practices. Experienced in working with datasets, performing data cleaning, and creating visual insights using tools like Power BI and Tableau. Familiar with building basic applications and implementing algorithm-based solutions. Actively learning advanced technologies and focused on developing practical, real-world projects to strengthen technical expertise and industry readiness.
Institute of Engineering and management kolkata
8
2023
-
2027
Honors & awards
Certificate of Paper Presentation (ISICVA 2025)
November 28, 2025
Handwritten Digit Recognition (HDR) is a complex task in Optical Character Recognition (OCR) due to variations in writing styles, stroke inconsistencies, and structural complexities of numerals. The recognition challenge intensifies for languages like Devnagari, Bengali and Assamese, where digit orientations and feature similarities contribute to ambiguity. Despite the success of deep learning in computer vision, HDR for these languages remains underexplored compared to English, Arabic, or Chinese. In this study, we propose a deep learning-based model for Devnagari, Bengali and Assamese handwritten digit recognition, leveraging a densely connected convolutional neural network (DenseNet). Unlike previous studies that rely on specific datasets, our approach is dataset-independent, ensuring robustness across various data sources. Advanced preprocessing and augmentation techniques are applied to enhance generalization. Our model has been systematically evaluated, demonstrating high accuracy in classifying handwritten numerals from both languages. Comparative analysis with prior works shows a significant reduction in error rates, highlighting the effectiveness of our approach. Furthermore, we provide an extensive review of existing challenges, recognition methodologies, and real-world applications in HDR, filling the research gap in Devnagari, Bengali and Assamese digit recognition. This research contributes to the development of more efficient and generalized HDR models, paving the way for future advancements in multilingual handwritten numeral recognition. In this paper, the accuracy for every training dataset in Devnagari, Bengali and Assamese is separately calculated which has the maximum accuracy 99.72%, 99%, 96.53%. Finally, 97.79% accuracy is obtained for Devnagari dataset, and 95.99% accuracy rate is obtained Bengali dataset and 96.38% in Assamese dataset.