 This paper investigates the use of convolutional neural networks, CNNs, for generating powerful feature representations for high-resolution remote sensing image retrieval, HRSR. First, a pre-trained CNN is used as a feature extractor since there are no sufficiently large remote sensing datasets to train a CNN from scratch. Then, a novel CNN architecture is proposed with fewer parameters than the pre-trained and fine-tuned CNNs. This CNN is able to learn low-dimensional features from limited labeled images. Finally, the proposed schemes are evaluated on several challenging datasets and show state-of-the-art performance. This article was authored by Weishu and Joe, Sean Newsom, So Minlee, and others.