Conditional Field for Image Classification

Abstract

We propose a novel multi label (ML) classification approach based on the Conditional Random fields (CRF) for the high resolution UAV images. The inderlying idea of the proposed model integrate 1) spatial information within the same class; jointly with 2) cross-correlation information between different class labels after . The experiments were done on two different UAV image datasets and the experimental results show that the new model outperforms conventional approaches.