Convolutional neural network (CNN) based methods have dominated the field of aerial scene classification for the past few years. While achieving remarkable success, CNN-based methods suffer from excessive parameters and notoriously rely on large amounts of training data. In this work, we introduce few-shot learning to the aerial scene classification problem. Few-shot learning aims to learn a model on base-set that can quickly adapt to unseen categories in novel-set, using only a few labeled samples. To this end, we proposed a meta-learning method for few-shot classification of aerial scene images. First, we train a feature extractor on all base categories to learn a representation of inputs. Then in the meta-training stage, the classifier is optimized in the metric space by cosine distance with a learnable scale parameter. At last, in the meta-testing stage, the query sample in the unseen category is predicted by the adapted classifier given a few support samples. We conduct extensive experiments on two challenging datasets: NWPU-RESISC45 and RSD46-WHU. The experimental results show that our method yields state-of-the-art performance. Furthermore, several ablation experiments are conducted to investigate the effects of dataset scale, the impact of different metrics and the number of support shots; the experiment results confirm that our model is specifically effective in few-shot settings.

In recent years, deep convolutional neural networks (CNNs) have demonstrated impressive ability to represent hyperspectral images (HSIs) and achieved encouraging results in HSI classification. However, the existing CNN-based models operate at the patch level, in which a pixel is separately classified into classes using a patch of images around it. This patch-level classification will lead to a large number of repeated calculations, and it is hard to identify the appropriate patch size that is beneficial to classification accuracy. In addition, the conventional CNN models operate convolutions with local receptive fields, which cause the failure of contextual spatial information modeling. To overcome these aforementioned limitations, we propose a novel end-to-end, pixel-to-pixel, fully convolutional spatial propagation network (FCSPN) for HSI classification. Our FCSPN consists of a 3-D fully convolution network (3D-FCN) and a convolutional spatial propagation network (CSPN). Specifically, the 3D-FCN is first introduced for reliable preliminary classification, in which a novel dual separable residual (DSR) unit is proposed to effectively capture spectral and spatial information simultaneously with fewer parameters. Moreover, the channel-wise attention mechanism is adapted in the 3D-FCN to grasp the most informative channels from redundant channel information. Finally, the CSPN is introduced to capture the spatial correlations of HSIs via learning a local linear spatial propagation, which allows maintaining the HSI spatial consistency and further refining the classification results. Experimental results on three HSI benchmark data sets demonstrate that the proposed FCSPN achieves state-of-the-art performance on HSI classification.

In order to remove speckle noise from original synthetic aperture radar (SAR) images effectively and efficiently, this paper proposes a hybrid dilated residual attention network (HDRANet) with residual learning for SAR despeckling. Firstly, HDRANet employs the hybrid dilated convolution (HDC) in lightweight network architecture to enlarge the receptive field and aggregate global information. Then, a simple yet effective attention module, convolutional block attention module (CBAM), is integrated into the proposed model to constitute a residual HDC attention block through skip connection, which further enhances representation power and performance of the model. Extensive experimental results on the synthetic and real SAR images demonstrate the superior performance of HDRANet over the state-of-the-art methods in terms of quantitative metrics and visual quality.

Onboard real-time object detection in remote sensing images is a crucial but challenging task in this computation-constrained scenario. This task not only requires the algorithm to yield excellent performance but also requests limited time and space complexity of the algorithm. However, previous convolutional neural networks (CNN) based object detectors for remote sensing images suffer from heavy computational cost, which hinders them from being deployed on satellites. Moreover, an onboard detector is desired to detect objects at vastly different scales. To address these issues, we proposed a lightweight one-stage multi-scale feature fusion detector called MSF-SNET for onboard real-time object detection of remote sensing images. Using lightweight SNET as the backbone network reduces the number of parameters and computational complexity. To strengthen the detection performance of small objects, three low-level features are extracted from the three stages of SNET respectively. In the detection part, another three convolutional layers are designed to further extract deep features with rich semantic information for large-scale object detection. To improve detection accuracy, the deep features and low-level features are fused to enhance the feature representation. Extensive experiments and comprehensive evaluations on the openly available NWPU VHR-10 dataset and DIOR dataset are conducted to evaluate the proposed method. Compared with other state-of-art detectors, the proposed detection framework has fewer parameters and calculations, while maintaining consistent accuracy.

Hyperspectral images (HSIs) provide abundant spectral and spatial information, playing an irreplaceable role in land-cover classification. Recently, based on deep learning (DL) technologies, an increasing number of HSI classification approaches have been proposed, which demonstrate promising performance. However, previous studies suffer from two major drawbacks: 1) the architecture of most DL models is manually designed, relies on specialized knowledge, and is relatively tedious. Moreover, in HSI classifications, datasets captured by different sensors have different physical properties. Correspondingly, different models need to be designed for different datasets, which further increases the workload of designing architectures and 2) the mainstream framework is a patch-to-pixel framework. The overlap regions of patches of adjacent pixels are calculated repeatedly, which increases computational cost and time cost. In addition, the classification accuracy is sensitive to the patch size, which is artificially set based on extensive investigation experiments. To overcome the issues mentioned above, we first propose a 3-D asymmetric neural network search algorithm and leverage it to automatically search for efficient architectures for HSI classifications. By analyzing the characteristics of HSIs, we specifically build a 3-D asymmetric decomposition search space, where spectral and spatial information is processed with different decomposition convolutions. Furthermore, we propose a new fast classification framework, i.e., pixel-to-pixel classification framework, which has no repetitive operations and reduces the overall cost. Experiments on three public HSI datasets captured by different sensors demonstrate the networks designed by our 3-D asymmetric neural architecture search (3-D-ANAS) achieve competitive performance compared to several state-of-the-art methods, while having a much faster inference speed. Code is available at: https://github.com/hkzhang91/3D-ANAS .