Recently, approaches based on deep learning are quite prevalent in the area of remote sensing scene classification. Though significant success has been achieved, these approaches are still subject to an excess of parameters and extremely dependent on a large quantity of labeled data. In this study, few-shot learning is used for remote sensing scene classification tasks. The goal of few-shot learning is to recognize unseen scene categories given extremely limited labeled samples. For this purpose, a novel task-adaptive embedding network is proposed to facilitate few-shot scene classification of remote sensing images, referred to as TAE-Net. A feature encoder is first trained on the base set to learn embedding features of input images in the pre-training phase. Then in the meta-training phase, a new task-adaptive attention module is designed to yield the task-specific attention, which can adaptively select informative embedding features among the whole task. In the end, in the meta-testing phase, the query image derived from the novel set is predicted by the meta-trained model with limited support images. Extensive experiments are carried out on three public remote sensing scene datasets: UC Merced, WHU-RS19, and NWPU-RESISC45. The experimental results illustrate that our proposed TAE-Net achieves new state-of-the-art performance for few-shot remote sensing scene classification.

Scene classification is a critical technology to solve the challenges of image search and image recognition. It has become an indispensable and challenging research topic in the field of remote sensing. At present, most scene classifications are solved by deep neural networks. However, existing methods require large-scale training samples and are not suitable for actual scenarios with only a few samples. For this reason, a framework based on metric learning and local descriptors (MLLD) is proposed to enhance the classification effect of remote sensing scenes on the basis of few-shot. Specifically, MLLD adopts task-level training that is carried out through meta-learning, and meta-knowledge is learned to improve the model’s ability to recognize different categories. Moreover, Manifold Mixup is introduced by MLLD as a feature processor for the hidden layer of deep neural networks to increase the low confidence space for smoother decision boundaries and simpler hidden layer representations. In the end, a learnable metric is introduced; the nearest category of the image is matched by measuring the similarity of local descriptors. Experiments are conducted on three public datasets: UC Merced, WHU-RS19, and NWPU-RESISC45. Experimental results show that the proposed scene classification method can achieve the most advanced results on limited datasets.

As a fundamental task in the field of remote sensing, scene classification is increasingly attracting attention. The most popular way to solve scene classification is to train a deep neural network with a large-scale remote sensing dataset. However, given a small amount of data, how to train a deep neural network with outstanding performance remains a challenge. Existing methods seek to take advantage of transfer knowledge or meta-knowledge to resolve the scene classification issue of remote sensing images with a handful of labeled samples while ignoring various class-irrelevant noises existing in scene features and the specificity of different tasks. For this reason, in this paper, an end-to-end graph neural network is presented to enhance the performance of scene classification in few-shot scenarios, referred to as the graph-based embedding smoothing network (GES-Net). Specifically, GES-Net adopts an unsupervised non-parametric regularizer, called embedding smoothing, to regularize embedding features. Embedding smoothing can capture high-order feature interactions in an unsupervised manner, which is adopted to remove undesired noises from embedding features and yields smoother embedding features. Moreover, instead of the traditional sample-level relation representation, GES-Net introduces a new task-level relation representation to construct the graph. The task-level relation representation can capture the relations between nodes from the perspective of the whole task rather than only between samples, which can highlight subtle differences between nodes and enhance the discrimination of the relations between nodes. Experimental results on three public remote sensing datasets, UC Merced, WHU-RS19, and NWPU-RESISC45, showed that the proposed GES-Net approach obtained state-of-the-art results in the settings of limited labeled samples.

Deep learning has a wide range of applications and far-reaching influence in our production and life, but the training of models requires a lot of data. In the real world, data acquisition requires a lot of costs, and the distribution of data is often uneven, with a small number of categories occupying a large number of samples, making the overall data present a long-tailed distribution. This makes convolutional neural network often perform poorly when training data are heavily class-imbalanced. In this work, enhance the feature extraction capabilities of the base model by add attention mechanism, and use the regularization technology mix-up algorithm to enhance the long-tail data. compared several state-of-the-arts techniques on the benchmark datasets imbalanced CIFAR10 and CIFAR100, that our method provides consistent and significant improvements over previous models.