Search

Automatic ship detection in SAR imagery has been playing a significant role in the field of marine monitoring but great challenges still exist in real-time application. Despite the exciting progresses made by deep-learning techniques, most detectors failed to yield locations of fairly high quality. Moreover, the ships with variant sizes and aspects are easily omitted especially for small objects under complicated background. To alleviate the above problem, the authors propose an elaborately designed single shot detection framework combined with attention mechanism, which roughly locates the regions of interest via an automatically learned attentional map. This lay the foundation of accurate positioning of extremely small objects since the background interference can be effectively suppressed. Furthermore, a multi-level feature fusion module integrated in top-down and bottom-up manner is adopted to adequately aggregate features from not only adjacent but also distant layers. This strengthens local details and merge strong semantic information, enabling the generation of higher qualified anchors for the efficient detection of multi-scale and multi-orientated objects. Experiments on SAR ship dataset have achieved a promising result, surpassing current state-of-the-art methods.

The proliferation of massive polarimetric Synthetic Aperture Radar (SAR) data helps promote the development of SAR image interpretation. Due to the advantages of powerful feature extraction capability and strong adaptability for different tasks, deep learning has been adopted in the work of SAR image interpretation and has achieved good results. However, most deep learning methods only employ single-polarization SAR images and ignore the water features embedded in multi-polarization SAR images. To fully exploit the dual-polarization SAR data and multi-scale features of SAR images, an effective flood detection method for SAR images is proposed in this paper. In the proposed flood detection method, a powerful Multi-Scale Deeplab (MS-Deeplab) model is constructed based on the dual-channel MobileNetV2 backbone and the classic DeeplabV3+ architecture to improve the ability of water feature extraction in SAR images. Firstly, the dual-channel feature extraction backbone based on the lightweight MobileNetV2 separately trains the dual-polarization SAR images, and the obtained training parameters are merged with the linear weighting to fuse dual-polarization water features. Given the multi-scale space information in SAR images, then, a multi-scale feature fusion module is introduced to effectively utilize multi-layer features and contextual information, which enhances the representation of water features. Finally, a joint loss function is constructed based on cross-entropy and a dice coefficient to deal with the imbalanced categorical distribution in the training dataset. The experimental results on the time series of Sentinel-1A SAR images show that the proposed method for flood detection has a strong ability to locate water boundaries and tiny water bodies in complex scenes. In terms of quantitative assessment, MS-Deeplab can achieve a better performance compared with other mainstream semantic segmentation models, including PSPNet, Unet and the original DeeplabV3+ model, with a 3.27% intersection over union (IoU) and 1.69% pixel accuracy (PA) improvement than the original DeeplabV3+ model. [ABSTRACT FROM AUTHOR]

At present, deep learning has been widely used in SAR ship target detection, but the accurate and real-time detection of multi-scale targets still faces tough challenges. CNN-based SAR ship detectors are challenged to meet real-time requirements because of a large number of parameters. In this paper, we propose a lightweight, single-stage SAR ship target detection model called YOLO-based lightweight multi-scale ship detector (LMSD-YOLO), with better multi-scale adaptation capabilities. The proposed LMSD-YOLO consists of depthwise separable convolution, batch normalization and activate or not (ACON) activation function (DBA) module, Mobilenet with stem block (S-Mobilenet) backbone module, depthwise adaptively spatial feature fusion (DSASFF) neck module and SCYLLA-IoU (SIoU) loss function. Firstly, the DBA module is proposed as a general lightweight convolution unit to construct the whole lightweight model. Secondly, the improved S-Mobilenet module is designed as the backbone feature extraction network to enhance feature extraction ability without adding additional calculations. Then, the DSASFF module is proposed to achieve adaptive fusion of multi-scale features with fewer parameters. Finally, the SIoU is used as the loss function to accelerate model convergence and improve detection accuracy. The effectiveness of the LMSD-YOLO is validated on the SSDD, HRSID and GFSDD datasets, respectively, and the experimental results show that our proposed model has a smaller model volume and higher detection accuracy, and can accurately detect multi-scale targets in more complex scenes. The model volume of LMSD-YOLO is only 7.6MB (52.77% of model size of YOLOv5s), the detection speed on the NVIDIA AGX Xavier development board reached 68.3 FPS (32.7 FPS higher than YOLOv5s detector), indicating that the LMSD-YOLO can be easily deployed to the mobile platform for real-time application. [ABSTRACT FROM AUTHOR]

Deep neural networks have achieved great progress in 3D scene understanding. However, recent methods mainly focused on objects with canonical orientations in contrast with random postures in reality. In this letter, we propose a hierarchical neural network, named Local Frame Network (LFNet), based on the local rotation invariant coordinate frame for robust point cloud analysis. The local point patches in different orientated objects are transformed into an identical distribution based on this coordinate frame, and the transformed coordinates are taken as input features to eliminate the influence of rotations at the input level. Meanwhile, a discrete convolution operator is defined in the constructed coordinate frame to extract rotation invariant features from local patches, which can further remove the influence of rotations at the convolution level. Moreover, a Spatial Feature Encoder (SFE) module is utilized to perceive the spatial structure of the local region. Mathematical analysis and experimental results on two public datasets demonstrate that the proposed method can eliminate the influence of rotations without data augmentation and outperforms other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Due to the ability of powerful feature representation, deep-learning-based object detection has attracted considerable research attention, and many methods have been proposed for remote sensing images. However, there are still some problems that need to be addressed. In this paper, a novel and effective detection framework based on faster region-based convolutional neural network is designed. Specifically, first, in order to locate the boundaries of large objects and find the missing small objects, DetNet is incorporated into the detection framework as the backbone network. DetNet fixes the spatial resolution in deep layers and adopts dilated bottleneck with convolution projection to increase the divergence between input and output feature maps. Then, the proposed framework uses the backbone network to extract the scene features and region features simultaneously, which are both mapped to feature vectors and then fused together. The feature fusion operation can improve the feature representation of the generated region. Last, to improve the performance of localization, the cascade structure is adopted in the framework. The cascade structure has multiple phases and every phase has independent classifier and regressor. The results obtained from the previous phase are used as the regions of interest in the next phase. Therefore, the multiphase detector can increase the detection accuracy phase by phase. Comprehensive evaluations on a public ten-class object detection dataset demonstrate the effectiveness of the proposed framework. Moreover, ablation experiments are also implemented to show the respective influence of different parts of the framework on the performance improvement. [ABSTRACT FROM AUTHOR]

The conventional convolutional neural network (CNN) has proven to be effective for synthetic aperture radar (SAR) target recognition. However, the relationship between different convolutional kernels is not taken into account. The lack of the relationship limits the feature extraction capability of the convolutional layer to a certain extent. To address this problem, this paper presents a novel method named weighted kernel CNN (WKCNN). WKCNN integrates a weighted kernel module (WKM) into the common CNN architecture. The WKM is proposed to model the interdependence between different kernels, and thus to improve the feature extraction capability of the convolutional layer. The WKM consists of variables and activations. The variable represents the weight of the convolutional kernel. The activation is a mapping function which is used to determine the range of the weight. To adjust the variable adaptively, back propagation (BP) algorithm for the WKM is derived. The training of the WKM is driven by optimizing the cost function according to the BP algorithm, and three training modes are presented and analysed. SAR target recognition experiments are conducted on the moving and stationary target acquisition and recognition (MSTAR) dataset, and the results show the superiority of the proposed method. [ABSTRACT FROM AUTHOR]

In the optical region, the global scattering center (SC) model is important for target recognition and data compression. However, the existing methods are difficult to be accomplished because of the requirement for large storage or time consumption. To solve this problem, a new easy-to-implement method is proposed in this paper. By employing a modified random sample consensus method, all candidate positions for the global SC model are extracted and discriminated efficiently. The clean step is unnecessary in the proposed method, which makes the procedure more efficient. Experimental results on the high-frequency-electromagnetic data demonstrate the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]