Research Projects [2021–Present]

My research specializes in Knowledge-Guided Machine Learning, Multimodal Understanding, Vision-Language Models (VLMs), and Remote Sensing for sustainable agriculture and infrastructure mapping. I focus on:

• Building large-scale benchmark datasets (IRRISIGHT, CAFOSat, IrrMap) integrating imagery, metadata, and text.
• Designing ML frameworks (IrrNet, IGraSS, KIIM) that combine graph constraints, domain knowledge, and multimodal fusion.
• Developing scalable pipelines for irrigation mapping and agricultural water management.

Undergraduate Thesis [2018-2019]

• REAL-TIME BANGLADESHI SIGN LANGUAGE DETECTION USING FASTER R-CNN:
  Accepted as a Full Paper in ICIET 2018. This is a real-time system which detects sign from a video and classifies the sign in the video. For the purpose of our work, we developed a robust dataset from scratch - BdSLImSet - which is available for further research. Our system is trained on Faster R-CNN and detects the sign with great accuracy. [Paper] [Dataset]

• BANGLADESHI SIGN WORD RECOGNITION USING LSTM:
  This is a system which detects sign words -sequence of actions- from a video. This works with sequential data and falls under the domain of action recognition. For the purpose of our work, we developed a robust dataset from scratch - BdSLVidSet - which is available for further research. This dataset is trained using CNN for feature extraction and LSTM to combine the sequential feature to recognize the words. Dataset

Undergraduate Course Projects [2015-2019]

• Efficient Noise Reduction System With Dataset:
  In this project, user can add noise - Salt Pepper Noise, Gaussian Noise and Speckle Noise – and apply filters Wiener Filter, Median Filter, Gaussian Filter by varying the parameters. (Tools: Matlab).
• Personality Analysis Using Various Feature Scaling On Dataset :
  It’s a human behavior analysis project using the data collected from our department. (Tools: Python and Data Analysis Libraries (NumPy, Panda, Matplotlib, Scikitlearn, TensorFlow, Keras)). [Code]
• E-COMMERCE WEBSITE WITH JAVA SERVLET AND RMI:
  E-commerce based website. (Tools: Java, Servlet, RMI, Bootstrap, Oracle 10g).[Code]
• PUZZLE GAME WITH A STAR ALGORITHM:
  This game is implemented using WebGL and JavaScript. (Tools: WebGL, JavaScript). [Code]
• SUPER-SHOP MANAGEMENT SYSTEM:
  Distributed database management project. (Tools: Oracle 10g, PL/SQL). [Code]
• ROBO MOS FACE RECOGNIZING MESSENGER:
  Server-client based face recognition and communication system. (Tool: Raspberry Pi, Python2, OpenCV, Pi Cam). [Code]
• SHOPEXPERT:
  E-commerce based website. (Tools: Visual Studio 2017, MVC, ASP.NET, C#). [Code]
• CINEFREAK:
  Movie listing based android app. (Tools: Android Studio, XML, Java, SQLite). [Code]
• BAZINGA:
  Movies information-based website with various user features. (Tools: HTML, PHP, MySQL, WAMP). [Code]
• SUP:
  Desktop based server-client chat application with file sharing option. (Tools: Java, Netbeans). [Code]
• WAR OF SPACESHIP:
  Desktop based shooting game with several fun features. (Tools: Visual Studio 2010, iGraphics, C). [Code]

Personal Projects [2018-2020]

• 37 Dogs Breed Classifier
• Face Recognition System Using Facenet (1:k matching problem)
• Sentiment Analysis from Facial Expression
• Neural Style Transfer
• Hand Segmentation (wearing gloves)
• Human Segmentation and Background Remove
• IMDB User Review Analysis