Distribution Consistency Loss for Large-Scale Remote Sensing Image Retrieval

Abstract

Remote sensing images are featured by massiveness, diversity and complexity. These features put forward higher requirements for the speed and accuracy of remote sensing image retrieval. The extraction method plays a key role in retrieving remote sensing images. Deep metric learning (DML) captures the semantic similarity information between data points by learning embedding in vector space. However, due to the uneven distribution of sample data in remote sensing image datasets, the pair-based loss currently used in DML is not suitable. To improve this, we propose a novel distribution consistency loss to solve this problem. First, we define a new way to mine samples by selecting five in-class hard samples and five inter-class hard samples to form an informative set. This method can make the network extract more useful information in a short time. Secondly, in order to avoid inaccurate feature extraction due to sample imbalance, we assign dynamic weight to the positive samples according to the ratio of the number of hard samples and easy samples in the class, and name the loss caused by the positive sample as the sample balance loss. We combine the sample balance of the positive samples with the ranking consistency of the negative samples to form our distribution consistency loss. Finally, we built an end-to-end fine-tuning network suitable for remote sensing image retrieval. We display comprehensive experimental results drawing on three remote sensing image datasets that are publicly available and show that our method achieves the state-of-the-art performance.

High-Resolution Remote Sensing Image Retrieval Based on Classification-Similarity Networks and Double Fusion

“…It should be stressed that some evaluation measures, such as P @k (see below), depend on the size of test set. Most empirical studies [1], [18], [19], [28], [31] use 20% of the dataset PatternNet for test, therefore, for purpose of a fair comparison, we randomly divide our PatternNet test set into four equal parts, compute evaluation measures over each part separately, and report the average values. Furthermore, we create image pairs in such a way that each class provides the same number of similar image pairs and each class combination provides the same number of dissimilar image pairs.…”

Learning to Hash with Convolutional Network for Multi-label Remote Sensing Image Retrieval

“…Image retrieval [9] explores and searches largescale database to access precise information efficiently. This search can be done using metadata or content-based data.…”

Improved ORB Algorithm Using Three-Patch Method and Local Gray Difference

“…A novel DML method named as global optimal structured loss to deal with the limitation of triplet loss [39]. To improve the problem that the pair-based loss currently used in DML is not appropriate, a novel distribution consistency loss was proposed in [40].…”