Projects
Machine Learning Based Event Participants Presence Prediction System with Facial Attendance
A machine learning system that predicts whether each event registrant will show up, then verifies it with webcam facial attendance. A Next.js and Node web app collects registrations in MongoDB, an XGBoost model served by FastAPI predicts Present or Absent, and an OpenCV face-recognition module marks who actually attended, so predicted and real attendance sit side by side.
Motion-Based Burglar Detection
A real-time burglar detector in Python that uses no neural network at all - just classic frame differencing in OpenCV. It compares two frames a few steps apart, boxes whatever moved, stamps a "Burglar Detected!" alert and saves a throttled snapshot to disk. The whole script is explained line by line.
Instance Segmentation with YOLOv8
A real-time instance-segmentation project in Python. YOLOv8's segmentation weights trace the exact pixel outline of each object in a video, and a built-in tracker keeps a stable ID on every instance across frames. The whole script is explained line by line.
Computer Vision with OpenCV & YOLOv8
A hands-on computer-vision journey in Python - from reading and drawing on images with OpenCV, through real-time object detection to object counting with YOLOv8. Every script is broken down line by line. (Instance segmentation and a motion-based burglar detector each get their own deep-dive in a separate project.)
Handwriting Recognition using PyTorch
A deep learning project that trains a Convolutional Neural Network (CNN) in PyTorch to recognise handwritten digits from the MNIST dataset, tracked end-to-end with TensorBoard.
Movie Recommendation System using TMDB Dataset
A content-based movie recommendation system using TF-IDF vectorization and cosine similarity on TMDB metadata.