Publications & Projects

• Ghosh, Ayanabha. (2023). Towards Full-page Offline Bangla Handwritten Text Recognition using Image-to-Sequence Architecture. 10.13140/RG.2.2.12455.73126. 

[Preprint link]

• A. Ghosh, "Sentiment Analysis of IMDb Movie Reviews : A comparative study on Performance of Hyperparameter-tuned Classication Algorithms," 2022 8th International Conference on Advanced Computing and Communication Systems (ICACCS), 2022, pp. 289-294, doi: 10.1109/ICACCS54159.2022.9784961.

[Publication link]

• Ghosh A., Das T., Majumder S., Roy A. (2020) Authentication of User in Connected Governance Model. In: Batra U., Roy N., Panda B. (eds) Data Science and Analytics. REDSET 2019. Communications in Computer and Information Science, vol 1230. Springer, Singapore. https://doi.org/10.1007/978-981-15-5830-6 10

[Publication link]

• Towards Full-page Offline Bangla Handwritten Text Recognition using Image-to-Sequence Architecture

  • • • Worked on this as my MTech Project/Thesis

      • Introduced a new dataset containing full-page Bangla handwritten scripts

      • Developed and applied an Image-to-Sequence Transformer architecture for full-page handwritten text recognition without prior segmentation

      • Accepted at IEEE SILCON 2023

      • Areas & Tech involved - Deep Learning, NLP, CNN, PyTorch, Transformers, Python 3

• Scene Reconstruction for Edge using Metadata

  • • • Worked on this as an Intern at Siemens RDA.

      • Performing Real-time Object detection and Pose estimation from video footage using pre-trained YOLO-v3 and MediaPipe models respectively.

      • Reconstructing or estimating the video feed at the Server end (Remote) using the extracted metadata.

      • Areas & Tech involved - Deep Learning (Applied), Computer Vision, OpenCV, Python 3, Onvif

      • Click here for details

• Sentiment Analysis of IMDb Movie Reviews : A study on Performance of Hyperparameter-tuned Classication Algorithms

  • • • Course project of Machine Learning (CSL7550) during Fall 2021

      • Published at IEEE ICACCS 2022 (See Publication Section)

      • Performing sentiment analysis of IMDb Movie reviews using three different supervised ML algorithms - Linear SVM, Naive Bayes, Logistic Regression Classier

      • Trained the models using different settings of their Hyperparameters

      • Evaluated them based on Accuracy score, F1 score and AUC scores

      • Areas & Tech involved - Machine Learning, NLP, Scikit-learn, Python 3, Information Retrieval

      • Click here for details

• Detection of Parkinson's disease from offline hand drawn Archimedean spiral

  • • • Worked on this during Summer School at ISI Kolkata in 2019

      • Developed a Deep learning image classication model using Keras to predict if a person is having Parkinson's disease or not

      • Dataset contained Hand-drawn Archimedean spirals drawn by different persons having or not having Parkinson's Disease

      • Areas & Tech involved - Deep Learning, Keras, CNN, Python 3

• Authentication of User in Connected Governance Model

  • • • Worked on this as my BTech Final year project

      • Published at REDSET 2019 (See Publication section)

      • Developed a 3-phase Authentication scheme for an already proposed Single window E-Governance model

      • Implemented a demo on a Banking service

      • Areas & Tech involved - Cryptography, Security, Core Java, NetBeans IDE, Socket Programming

• Several other ML/DL projects are available on my GitHub Repository.