The Bank Customer Churn Prediction project aims to identify customers likely to leave the bank, enabling proactive retention strategies. An Artificial Neural Network model was implemented using a Sequential neural network with two hidden layers to classify customers as "churning" or "non-churning.

Designed and implemented a real-time news recommendation system using DQN and A2C on the MIND dataset. Utilized vector embedding (TF-IDF, GloVe) and context-aware features for user-article representation, enabling accurate content personalization and long-term engagement prediction.

A real-time parking occupancy detection system built with OpenCV and Machine Learning (SVM). The project processes parking lot surveillance video, detects whether each parking spot is occupied or vacant, and overlays results directly on the live video feed.

The Finger Count project is a computer vision application that uses OpenCV and MediaPipe to detect and count the number of fingers raised by a hand in real-time. OpenCV is used for preprocessing the video frames, such as resizing, drawing landmarks, and displaying the output.

Developed a Facial Emotion Recognition (FER) system on the FER-2013 dataset to classify emotions, addressing challenges of low resolution and noisy facial images. Trained and optimized Convolutional Neural Network (CNN) models and fine-tuned pretrained architectures to achieve improved classification accuracy.

The Face, Eyes, and Palm Detection project is a computer vision application that uses Haar Cascade Classifiers to detect faces, eyes, and palms in real-time video streams or static images. Haar cascades are pre-trained classifiers that scan images at multiple scales to detect objects based on patterns of features like edges and textures.