Yunjie Li

Yunjie Li

Auto-generated profile, based on publicly available information

Yunjie Li

School of Information and Electronics, Beijing Institute of Technology, Beijing, China

Research Focus & Metrics

Academic Career Length

14Years (2013-Present)

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

Representative Publications

View All Publications
1 April 2020 in Signal Processing
Verified
Journal Article
Closed access
MZMengtao Zhu
Mengtao Zhu
YLYunjie Li
Yunjie Li
ZPZesi Pan
Zesi Pan
JYJian Yang
Jian Yang

24 July 2020 in IET Radar, Sonar & Navigation
Verified
Journal Article
Open access
PDF file
YLYunjie Li
Yunjie Li
MZMengtao Zhu
Mengtao Zhu
YMYihao Ma
Yihao Ma
JYJian Yang
Jian Yang
Recognition of multi‐function radar (MFR) work mode in an input pulse sequence is a fundamental task to interpret the functions and behaviour of an MFR. There are three major challenges that must be addressed: (i) The received radar pulses stream may contain an unknown number of multiple work mode class segments. (ii) The intra‐mode and inter‐mode knowledge of a modern MFR may be too flexible and complicated to be represented and learned through traditional hand‐crafted features and learning models. (iii) The variable duration of each enclosed work mode makes the identification of the transition boundaries of adjacent modes difficult. To address these challenges and implement automatic recognition of MFR work mode sequences at a pulse‐level, this study develops a novel processing framework based on a time series representation of MFR work mode sequence and sequence‐to‐sequence (seq2seq) long short‐term memory network. The proposed method can not only automatically recognise multiple complexes modulated work mode classes in a pulse sequence. Still, it can also accurately identify the transition boundaries between each class by labelling the class information for each pulse. The experimental results showed the extended capabilities and improved performance of the proposed method over the state‐of‐the‐art work mode classification methods.

19 October 2023 in IEEE Transactions on Intelligent Vehicles
Verified
Journal Article
Closed access
LFLili Fan
Lili Fan
DWDandan Wang
Dandan Wang
JWJunhao Wang
Junhao Wang
YLYunjie Li
Yunjie Li
YCYifeng Cao
Yifeng Cao
YWYutong Wang
Yutong Wang
Pavement defect detection is of profound significance regarding road safety, so it has been a trending research topic. In the past years, deep learning based methods have turned into a key technology, with advantages of high accuracy, strong robustness, and adaptability to complex pavement environments. This paper first reviews the methods based on image processing and 3D imaging. As for image-based processing techniques, they can be classified into three types based on how to label the collected data: fully supervised learning, unsupervised learning, and other methods. Different methods are further classified and compared with each other. Second, the pavement detection methods based on 3D data are sorted out, thereby summarizing their benefits, drawbacks, and application scenarios. Third, the study proposed the major challenges in the field of pavement defect detection, introduced validated datasets and evaluation metrics. Finally, on the basis of reviewing the literature in pavement defect detection, the promising direction is put forward.

15 July 2020 in IEEE Signal Processing Letters
Verified
Journal Article
Closed access
ZPZesi Pan
Zesi Pan
SWShafei Wang
Shafei Wang
MZMengtao Zhu
Mengtao Zhu
YLYunjie Li
Yunjie Li
In an ever-increasingly complex electromagnetic environment, multiple low probability of intercept (LPI) radar emitters may transmit their own signals simultaneously on similar bands, resulting in overlapping receiving signals in both time and frequency domain. In this letter, a novel Multi-Instance MultiLabel learning framework based on Deep Convolutional Neural Network (MIML-DCNN) is proposed to automatically recognize the overlapping LPI radar signals,which is trained by single type of signals only. The framework handles signals in an end-to-end manner that is integrated with a well-designed instance generation module, a sophisticated MIML classifier, and an adaptive threshold calibration. Through comprehensive experiments on simulated overlapping signals with four different modulation types, we prove that the proposed framework identifies each individual signal type precisely in the presence of overlapping signals, and is also robust to variation of the signal-to-noise ratio (SNR) and power ratio conditions.

26 October 2021 in IEEE Transactions on Aerospace and Electronic Systems
Verified
Journal Article
Closed access
MZMengtao Zhu
Mengtao Zhu
SWShafei Wang
Shafei Wang
YLYunjie Li
Yunjie Li
Deinterleaving mixtures of radar pulse sequences is the first and the most vital step for modern electronic reconnaissance systems to intercept and analyze the intentions of noncooperative radars. Currently, these systems are facing great challenges due to the emerging complexity of radar modulations. This article proposes a novel method for deinterleaving mixtures of radar pulse sequences based on the time series characteristics of each source (component pulse sequences). First, the mathematical representations are established to describe the structural characteristics of each source. Then, the deinterleaving problem is formulated as a minimization of a maximum-likelihood cost function that can be solved efficiently through a supervised neural machine translation (NMT) network, i.e., to translate each received pulse in the interleaved pulse sequence to the corresponding source label. A sequence-to-sequence NMT model is proposed to capture the structural relationships among the nonadjacent pulses originated from the same source in the mixtures and assign the corresponding label to each pulse. The proposed method does not require the exact knowledge of each component pulse sequence. The experimental results based on the time of arrival sequence of mixed pulse sequences show that the proposed method outperforms the state-of-the-art deinterleaving methods and achieves satisfactory performance under highly nonideal situations with measuring noise and lost pulse conditions. The proposed method can be directly applied to other fields involving deinterleaving problems and multivariate input conditions.

1 November 2023 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
LFLili Fan
Lili Fan
YZYu Zhou
Yu Zhou
HLHongmei Liu
Hongmei Liu
YLYunjie Li
Yunjie Li
DCDongpu Cao
Dongpu Cao
Remote sensing semantic segmentation plays a significant role in various applications such as environmental monitoring, land use planning, and disaster response. CNNs have been dominating remote sensing semantic segmentation. However, due to the limitations of convolution operations, CNNs cannot effectively model global context. The success of Transformers in the NLP domain provides a new solution for global context modeling. Inspired by Swin Transformer, we propose a novel remote sensing semantic segmentation model called CSTUNet. This model employs a dual-encoder structure consisting of a CNN-based main encoder and a Swin Transformer-based auxiliary encoder. We first utilize a detail-structure preservation module (DPM) to mitigate the loss of detail and structure information caused by Swin Transformer downsampling. Then we introduce a spatial feature enhancement module (SFE) to collect contextual information from different spatial dimensions. Finally, we construct a position-aware attention fusion module (PAFM) to fuse contextual and local information. Our proposed model obtained 70.75% MIoU on the ISPRS-Vaihingen dataset and 77.27% MIoU on the ISPRS-Potsdam dataset.