Learning to Assign Orientations to Feature Points
This work addresses feature point matching in computer vision, offering a generalizable solution that is incremental but enhances existing methods.
The paper tackles the problem of assigning canonical orientations to feature points for improved matching by training a Convolutional Neural Network with Siamese networks and a novel activation function, achieving state-of-the-art performance across multiple datasets without retraining.
We show how to train a Convolutional Neural Network to assign a canonical orientation to feature points given an image patch centered on the feature point. Our method improves feature point matching upon the state-of-the art and can be used in conjunction with any existing rotation sensitive descriptors. To avoid the tedious and almost impossible task of finding a target orientation to learn, we propose to use Siamese networks which implicitly find the optimal orientations during training. We also propose a new type of activation function for Neural Networks that generalizes the popular ReLU, maxout, and PReLU activation functions. This novel activation performs better for our task. We validate the effectiveness of our method extensively with four existing datasets, including two non-planar datasets, as well as our own dataset. We show that we outperform the state-of-the-art without the need of retraining for each dataset.