Seonglae ChoProposes a regularization term that encourages embeddings (local feature descriptors) to spread out evenly across the entire space. When combined with existing distance/triplet losses, it increases representation diversity and discriminative power
Learning Spread-Out Local Feature Descriptors
Learning Spread-out Local Feature Descriptors
We propose a simple, yet powerful regularization technique that can be used to significantly improve both the pairwise and triplet losses in learning local feature descriptors. The idea is that in...
https://arxiv.org/abs/1708.06320
