Wei Dai

Wei Dai

Auto-generated profile, based on publicly available information

Wei Dai

Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing, China

Research Focus & Metrics

Academic Career Length

24Years (2003-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
8 January 2024 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
DPDece Pan
Dece Pan
XGXin Gao
Xin Gao
WDWei Dai
Wei Dai
JFJiamei Fu
Jiamei Fu
ZWZhirui Wang
Zhirui Wang
Youming WuYouming Wu
Youming Wu
Synthetic aperture radar (SAR) ship detection plays an important role in the field of maritime security. However, certain unique imaging properties make it challenging to extract the shape features of ships, such as speckle noise and strong scattering interference from irrelevant objects. These factors result in inaccurate ship localization and obvious false alarms under complex large-scale inshore scenes. To address this issue, we propose the scattering region topology network (SRT-Net), which can dynamically capture the comprehensive global context and enhance the ship saliency. This is achieved through two key modules, namely the scattering region topological structure pyramid (SRTP) and the ship saliency enhancement (SSE) module. The former provides richer semantic information to distinguish the object from the background, while the latter offers an extra pixel-level classification task to guide accurate bounding box regression. Thanks to the guidance of richer information, the proposed method can achieve fewer false alarms and enhance location accuracy. Additionally, we introduce a scale feature adaptive (SFA) loss to balance the attention to ships with various scales, which improves the robustness of multiscale ship detection. The proposed method achieves state-of-the-art performance under complex inshore scenes, and its effectiveness is verified by experiments on a large-scale SAR ship detection dataset (LSSDD) and a public SAR ship detection dataset (SSDD+).

18 December 2024 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
Youming WuYouming Wu
Youming Wu
YSYuxi Suo
Yuxi Suo
QMQingbiao Meng
Qingbiao Meng
WDWei Dai
Wei Dai
TMTian Miao
Tian Miao
XSXian Sun
Xian Sun
Object detection and recognition (OD&R) based on deep learning is a hot topic in the application of synthetic aperture radar (SAR). These methodologies based on deep learning are inherently data-driven, which means that their performance is subjected to the corresponding datasets. Although existing datasets have included some common targets collected from real-valued intensity SAR images, they still exist some limitations in terms of quantity, categories, diversities, data domain. Hence, it is urgent to establish a large-quantity benchmark for fine-grained OD&R on complex-valued SAR images, which contains rich signal-domain features well coupled with classical physical modeling. Additionally, considering the unique imaging characteristics and diverse imaging conditions, some important attribute information, such as incidence and attitude angles, is necessary to be attached. In this paper, we propose a novel benchmark dataset with more than 340k instances for Fine-grAIned object detection and Recognition based on single look Complex SAR images, which is named as FAIR-CSAR. We collected complex-valued SAR images with a resolution of 1m to 5m from 175 entire images of Gaofen-3 covering 32 cities and multiple sea areas worldwide. All instances in the FAIR-CSAR are annotated by oriented bounding boxes (OBB), covering 5 major categories and 22 subcategories. Compared with existing datasets dedicated to OD&R, the FAIR-CSAR dataset has four particular advantages: (1) it contains complex-valued SAR images from various acquisition modes and polarization modes, including full-scale signal-domain features for object recognition; (2) it is much larger than other existing OD&R datasets in terms of quantity of instances; (3) it provides more fine-grained category annotation, and more detailed attribute information; (4) it provides more challenging images with some common imaging phenomena, such as speckle noise, azimuth ambiguities and so on. To establish a baseline adapted for single look complex SAR images, a multi-domain feature extraction and fusion network (MDNet) is proposed as a novel framework to mine detailed information underlying in various domains. And a series of state-of-the-art (SOTA) algorithms are applied on the FAIR-CSAR to build the fine-grained OD&R benchmark. Experimental results indicate that FAIR-CSAR is closer to practical application and more challenging than existing datasets for SAR images.

6 January 2021 in IEEE Access
Verified
Journal Article
Open access
WDWei Dai
Wei Dai
WDWenhui Diao
Wenhui Diao
XSXian Sun
Xian Sun
YZYue Zhang
Yue Zhang
ZLzhao liangjin
zhao liangjin
KFKun Fu
Kun Fu
Open Set Fine-Grained Recognition (OSFGR) aims at distinguishing known fine-grained categories from the data (contain known and unknown categories). The feature distribution of the known fine-grained category usually has the characteristics of small inter-class variance and large intra-class variance. Directly using the traditional Open Set Recognition (OSR) method on OSFGR does not achieve competitive performance. This is mainly due to the fact that the traditional OSR method is designed based on SoftMax function, whose translation invariance for input weakens the representation ability for the known fine-grained category in OSFGR. To settle this problem, we present a unified method based on class activation mapping value (CAMV) for OSFGR, which preserves the original discriminate feature. Besides, most OSR methods are based on the discrimination model with multi-class thresholds. When most known categories need to recognize, it will face the high computational cost and low efficiency. To overcome this problem, we propose the discrimination model with fusion one-class thresholds, which can greatly improve the inference speed without losing the recognition accuracy. Meanwhile, our idea can settle the extreme situation of OSR (only one known category). Our method achieves the state-of-art performance on the fine-grained open set datasets consisting of 4 public fine-grained datasets.

30 December 2024 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
QMQingbiao Meng
Qingbiao Meng
Youming WuYouming Wu
Youming Wu
YSYuxi Suo
Yuxi Suo
WDWei Dai
Wei Dai
ZYZhiyuan Yan
Zhiyuan Yan
XSXian Sun
Xian Sun
Aircraft detection in synthetic aperture radar (SAR) imagery is significant due to its critical role in various applications, including surveillance, reconnaissance, and security. However, given the background interference and discreteness of aircraft scattering, detectors are prone to acquire unremarkable aircraft features. These factors lead to false alarms and present difficulties in locating aircraft accurately. This article proposes an innovative scattering topology cue-based network (STC-Net), which enhances aircraft discriminability and more accurately evaluates the quality of the prediction results. We model the aircraft with the star topology (ST), which not only emphasizes critical components like the nose and wings but also explicitly links them as a cohesive unit. Based on the cue of ST, the star topology space fusion module (ST-SFM) and the star topology channel attention module (ST-CAM) are designed. The former integrates discrete components to reestablish the aircraft features based on neighboring information of ST, while the latter suppresses background interference to highlight the aircraft by exploiting node information of ST. In addition, completeness and consistency loss (CCLoss) function that includes the completeness-aware label and the positive sample weighting function is introduced. The completeness-aware label describe the localization accuracy by incorporating the degree of overlap of predicted results on ST, while the positive sample weighting function enhances the consistency of the classification and localization branches. Furthermore, experiments conducted on the Gaofen-3 SAR Aircraft Detection Dataset (GF3ADD) and the publicly available SAR-AIRcraft-1.0 dataset demonstrate the effectiveness and generalizability of STC-Net, with our method achieving state-of-the-art performance.

11 August 2022 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
JYJinze Yang
Jinze Yang
KFKun Fu
Kun Fu
Youming WuYouming Wu
Youming Wu
WDWenhui Diao
Wenhui Diao
WDWei Dai
Wei Dai
XSXian Sun
Xian Sun
The resolution degradation poses a huge challenge for object detection (OD) in aerial imagery. The existing methods use super-resolution (SR) based on generative adversarial network (GAN) to restore texture details in degraded images. However, constrained detection results are still acquired due to object feature difference between restored and clear images. Therefore, we propose a simple yet effective learning method called mutual-feed learning (MFL) to solve the problem in this article. A closed-loop structure is designed via building the feedback connection based on the feedforward connection between the two tasks. It effectively delivers the object spatial and feature information from OD to SR and provides restoration-enhanced images from SR to OD. Specifically, a feedback of region of interest (FROI) module is introduced to realize a region-level discrimination under the guidance of object information. It guides the discrimination process of SR. Furthermore, a multiscale object information (MSOI) module is developed to implement a feature-level restoration by narrowing the differences in object-related features. It improves the generation process of SR. Then OD can be performed in restoration-enhanced images to obtain more accurate results. Extensive experiments over Northwestern Polytechnical University (NWPU) Very High Resolution (VHR)-10, Cars Overhead With Context (COWC), and Fine-grained Object Recognition (FAIR)1M datasets show that the method can achieve state-of-the-art results.

19 September 2023 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
JCJinyue Chen
Jinyue Chen
Youming WuYouming Wu
Youming Wu
XGXin Gao
Xin Gao
WDWei Dai
Wei Dai
XZXuan Zeng
Xuan Zeng
XSXian Sun
Xian Sun
Intelligent ship detection algorithms for synthetic aperture radar (SAR) images have achieved significant results in Earth observation applications. By learning features such as scale, shape and texture from samples, they can quickly locate and recognize ships in complex backgrounds. However, due to the lack of use of polarization features, the upper bound of detection performance is still limited, especially under poor image quality conditions such as ambiguous interference. To solve this, the dual-polarization image feature fusion network (DPFF-Net) is proposed. The key of it lies in adaptive mining, enhancement and fusion of polarization features through the designed siamese structure, polarization-aware enhancement block (PAEB) and dynamic gated fusion block (DGFB). With fully utilizing complementary information hidden between co-polarization and cross-polarization data, more comprehensive and accurate features are obtained and used as the detect head input. Thus, the proposed algorithm achieves state-of-the-art performance, and its effectiveness are validated by experiments on dual-polarization SAR datasets.