Projects

This project introduces a simulation framework that transforms 2D RGB images into interactive physics-based objects. Leveraging pre-trained PyTorch models for object detection and depth estimation, combined with Panda3D for scene reconstruction and a custom algorithm to search for 3D models, the framework offers a user-friendly interface for selecting and integrating 3D representations into the simulated environment.

This project aims to develop a machine-learning model for predicting song genres based on lyrics. By leveraging natural language processing and various types of Neural networks (LSTM, CNN, etc.), the system seeks to enhance genre classification accuracy, and provide valuable insights into patterns of particular genres.