Efficient and Accurate Visual Relocalization with Sparse Neural Radiance Fields

An efficient and accurate framework for visual relocalization using a hybrid representation of an explicit geometric map and an implicit learning map with sparse neural radiance fields.