Tracking object across video frames, whether in a monocular or multiview scenario, is crucial for various applications ranging from surveillance to augmented reality. This project aims to advance the field of object tracking by employing sophisticated decision algorithms like Monte Carlo Tree Search (MCTS) and its reinforcement learning variant, AlphaZero, to enhance the association of people detection across different frames or views.
Project Overview
The goal of this project is to develop a robust object tracking system that can accurately associate detections across a series of video frames. This involves dealing with challenges such as occlusions, varying lighting conditions, and changes in perspective, especially in multiview setups.
The first step will be to implement and evaluate the performance of Monte Carlo Tree Search and AlphaZero in the context of people tracking.
Prerequisites
Participants in this project should have:
– Python programming experience
– Previous experience in machine learning and computer vision (Pytorch).
Contact
Please send an email to [email protected] if you are interested in this project.