Open vocabulary VQA for satellite imagery

Abstract

Methods and systems for question answering include generating (212) a tile from an image and encoding (214) the tile to generate an embedding vector. A set of neighbor objects is generated (302) based on an object of interest from a query. A first similarity is determined (306) between the object of interest and the embedding vector of the tile. Second similarities are determined (308) between the neighbor objects and the embedding vector of the tile. It is determined (316) that the tile is relevant responsive to the first similarity being greater than the second similarities. The query is answered (230) using the tile.

Images