This page highlights my research and development projects in Artificial Intelligence, Computer Vision, and Full-Stack AI applications, ranging from my PhD work to my Master's thesis and beyond.

Research & Major Projects

CNN-LSTM-based Activity Detection System for Construction Sites

Type: Computer Vision Project
Tools: Python, CNN, LSTM
Link: View on GitHub

  • Developed a computer vision-based activity recognition system specifically for construction environments.
  • Combined Convolutional Neural Networks (CNN) for spatial features with Long Short-Term Memory (LSTM) networks for temporal action sequence recognition.

Deepfake Detection (MS Thesis)

Type: Master's Thesis
Tools: Python, FaceSwapGAN, Landmark Detection

  • Developed a robust deepfake detection system by analyzing lighting inconsistencies in retinal and facial movements.
  • Implemented precise facial landmark detection to identify biometric anomalies.
  • Utilized FaceSwapGAN architectures to evaluate and improve detection accuracy against generative models.

Smart YouTube Video Assistant

Type: Generative AI / RAG
Tools: Python, LLMs, NLP
Link: View on GitHub

  • Developed an AI assistant that processes YouTube URLs and custom prompts to generate multi-modal responses (text, images, or video snippets).
  • Built a summarization pipeline to enhance user experience through contextual recommendations and related web sources.

Intelligent Health Monitoring System (FYP BS)

Type: Bachelor's Final Year Project
Tools: Django, Python, PostgreSQL, High Charts API
Link: View on GitHub

  • Implemented a comprehensive health monitoring dashboard featuring predictive forecasting.
  • Integrated a regression-based backend with a dynamic frontend using Bootstrap and JavaScript.
  • Handled full-stack responsibilities including database schema design and AI-driven health analytics.

Additional AI & Computer Vision Projects

Automated Stationery Calculator using YOLOv5

  • Developed an AI-based automatic billing system for retail stationery using the YOLOv5 object detection algorithm.
  • Optimized image processing pipelines in Python for real-time item identification and price calculation.

Image Steganography using Deep Learning

  • Implemented an advanced framework for the enhancement of encoded covers in multi-image steganography.
  • Leveraged deep learning frameworks to maximize data hiding capacity while maintaining visual imperceptibility.

Other Technical Projects

  • Content-Based Image Retrieval: Searching image databases based on visual features.
  • Font Identification: Using ML to classify and identify typography from images.
  • Genetic Algorithm Schedule Generator: Optimizing exam schedules using evolutionary computing.
  • Facial Emotion Recognition: Real-time classification of human emotions.
  • Dementia Classification: AI-driven diagnostic assistance for neurodegenerative conditions.

Technical Stack

  • Languages: Python, PHP, SQL, C, C++, JavaScript
  • AI Frameworks: TensorFlow, PyTorch, YOLOv8/v5, Scikit-learn, NLTK, MediaPipe
  • Backend & Web: Django, HTML5, CSS, Bootstrap, PostgreSQL
  • Tools: Git, Jupyter Notebook, Google Colab, Overleaf, Tableau, OpenCV