The goal of this project is to create a local feature matching algorithm using techniques described in Szeliski chapter 4.1. The pipeline we suggest is based on a simplified version of the famous SIFT pipeline. However, we will not be implementing this in the classical fashion. We will be implementing it as though it were part of a neural network. The matching pipeline is intended to work for instance-level matching – multiple views of the same physical scene.
Project Design Attribution: Georgia Institute of Technology, CS 6476 (Fall 2020), Frank Dellaert