Search

Your search keyword '"Mu, Wenjie"' showing total 29 results
Start Over Author "Mu, Wenjie"
29 results on '"Mu, Wenjie"'

1. VNI-Net: Vector Neurons-based Rotation-Invariant Descriptor for LiDAR Place Recognition

Author

Tian, Gengxuan, Zhao, Junqiao, Cai, Yingfeng, Zhang, Fenglin, Mu, Wenjie, and Ye, Chen

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

LiDAR-based place recognition plays a crucial role in Simultaneous Localization and Mapping (SLAM) and LiDAR localization. Despite the emergence of various deep learning-based and hand-crafting-based methods, rotation-induced place recognition failure remains a critical challenge. Existing studies address this limitation through specific training strategies or network structures. However, the former does not produce satisfactory results, while the latter focuses mainly on the reduced problem of SO(2) rotation invariance. Methods targeting SO(3) rotation invariance suffer from limitations in discrimination capability. In this paper, we propose a new method that employs Vector Neurons Network (VNN) to achieve SO(3) rotation invariance. We first extract rotation-equivariant features from neighboring points and map low-dimensional features to a high-dimensional space through VNN. Afterwards, we calculate the Euclidean and Cosine distance in the rotation-equivariant feature space as rotation-invariant feature descriptors. Finally, we aggregate the features using GeM pooling to obtain global descriptors. To address the significant information loss when formulating rotation-invariant descriptors, we propose computing distances between features at different layers within the Euclidean space neighborhood. This greatly improves the discriminability of the point cloud descriptors while ensuring computational efficiency. Experimental results on public datasets show that our approach significantly outperforms other baseline methods implementing rotation invariance, while achieving comparable results with current state-of-the-art place recognition methods that do not consider rotation issues.

Published
2023

2. Learning Sequence Descriptor based on Spatio-Temporal Attention for Visual Place Recognition

Author

Zhao, Junqiao, Zhang, Fenglin, Cai, Yingfeng, Tian, Gengxuan, Mu, Wenjie, Ye, Chen, and Feng, Tiantian

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Visual Place Recognition (VPR) aims to retrieve frames from a geotagged database that are located at the same place as the query frame. To improve the robustness of VPR in perceptually aliasing scenarios, sequence-based VPR methods are proposed. These methods are either based on matching between frame sequences or extracting sequence descriptors for direct retrieval. However, the former is usually based on the assumption of constant velocity, which is difficult to hold in practice, and is computationally expensive and subject to sequence length. Although the latter overcomes these problems, existing sequence descriptors are constructed by aggregating features of multiple frames only, without interaction on temporal information, and thus cannot obtain descriptors with spatio-temporal discrimination.In this paper, we propose a sequence descriptor that effectively incorporates spatio-temporal information. Specifically, spatial attention within the same frame is utilized to learn spatial feature patterns, while attention in corresponding local regions of different frames is utilized to learn the persistence or change of features over time. We use a sliding window to control the temporal range of attention and use relative positional encoding to construct sequential relationships between different features. This allows our descriptors to capture the intrinsic dynamics in a sequence of frames.Comprehensive experiments on challenging benchmark datasets show that the proposed approach outperforms recent state-of-the-art methods.The code is available at https://github.com/tiev-tongji/Spatio-Temporal-SeqVPR., Comment: 8 pages, 6 figures, published to RA-L

Published
2023
Full Text
View/download PDF

16. Learning Sequence Descriptor based on Spatiotemporal Attention for Visual Place Recognition

Author

Zhang, Fenglin, Zhao, Junqiao, Cai, Yingfeng, Tian, Gengxuan, Mu, Wenjie, and Ye, Chen

Subjects
FOS: Computer and information sciences, Computer Vision and Pattern Recognition (cs.CV), Computer Science - Computer Vision and Pattern Recognition
Abstract

Visual Place Recognition (VPR) aims to retrieve frames from a geotagged database that are located at the same place as the query frame. To improve the robustness of VPR in perceptually aliasing scenarios, sequence-based VPR methods are proposed. These methods are either based on matching between frame sequences or extracting sequence descriptors for direct retrieval. However, the former is usually based on the assumption of constant velocity, which is difficult to hold in practice, and is computationally expensive and subject to sequence length. Although the latter overcomes these problems, existing sequence descriptors are constructed by aggregating features of multiple frames only, without interaction on temporal information, and thus cannot obtain descriptors with spatio-temporal discrimination. In this paper, we propose a sequence descriptor that effectively incorporates spatio-temporal information. Specifically, spatial attention within the same frame is utilized to learn spatial feature patterns, while attention in corresponding local regions of different frames is utilized to learn the persistence or change of features over time. We use a sliding window to control the temporal range of attention and use relative position encoding to construct sequential relationships between different features. This allows our descriptors to capture the intrinsic dynamics in a sequence of frames. Comprehensive experiments on challenging benchmark datasets show that the proposed approach outperforms recent state-of-the-art methods.

Published
2023

19. Multi‐omics reveal differentiation and maintenance of dimorphic flowers in an alpine plant on the Qinghai‐Tibet Plateau.

Author

Zhu, Mingjia, Wang, Zhenyue, Yang, Yongzhi, Wang, Zefu, Mu, Wenjie, and Liu, Jianquan

Subjects
MOUNTAIN plants, FLOWERING of plants, MULTIOMICS, ANGIOSPERMS, FLOWER development, POLLINATION, CAROTENOIDS
Abstract

Dimorphic flowers growing on a single individual plant play a critical role in extreme adaption and reproductive assurance in plants and have high ecological and evolutionary significance. However, the omics bases underlying such a differentiation and maintenance remain largely unknown. We aimed to investigate this through genomic, transcriptome and metabolomic analyses of dimorphic flowers in an alpine biennial, Sinoswertia tetraptera (Gentianaceae). A high‐quality chromosome‐level genome sequence (903 Mb) was first assembled for S. tetraptera with 31,359 protein‐coding genes annotated. Two rounds of recent independent whole‐genome duplication (WGD) were revealed. Numerous genes from the recent species‐specific WGD were found to be differentially expressed in the two types of flowers, and this may have helped contribute to the origin of this innovative trait. The genes with contrasting expressions between flowers were related to biosynthesis of hormones, floral pigments (carotenoids and flavonoids) and iridoid compounds, which are involved in both flower development and colour. Metabolomic analyses similarly suggested differential concentrations of these chemicals in the two types of flowers. The expression interactions between multiple genes may together lead to contrasting morphology and chemical concentration and open versus closed pollination of the dimorphic flowers in this species for reproductive assurance. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

21. Early diversification and karyotype evolution of flowering plants

Author

Sun, Pengchuan, primary, Yang, Yongzhi, additional, Liu, Liang, additional, Xi, Zhenxiang, additional, Lin, Hao, additional, Feng, Landi, additional, Ma, Jianxiang, additional, Hu, Hongyin, additional, Bi, Guiqi, additional, Jiao, BeiBei, additional, Wang, Shang, additional, Yang, Jiao, additional, Mu, Wenjie, additional, Zhu, Mingjia, additional, Wang, Xiyin, additional, Wang, Zhenyue, additional, Davis, Charles, additional, and Liu, Jianquan, additional

Published
2022
Full Text
View/download PDF

26. A chromosome‐level genome assembly for the tertiary relict plant Tetracentron sinense oliv. (trochodendraceae)

Author

Li, Minjie, primary, Yang, Yongzhi, additional, Xu, Renping, additional, Mu, Wenjie, additional, Li, Ying, additional, Mao, Xingxing, additional, Zheng, Zeyu, additional, Bi, Hao, additional, Hao, Guoqian, additional, Li, Xiaojie, additional, Xu, Xiaoting, additional, Xi, Zhenxiang, additional, Shrestha, Nawal, additional, and Liu, Jianquan, additional

Published
2021
Full Text
View/download PDF

27. Genomes and demographic histories of the endangered Bretschneidera sinensis (Akaniaceae).

Author

Zhang, Han, Du, Xin, Dong, Congcong, Zheng, Zeyu, Mu, Wenjie, Zhu, Mingjia, Yang, Yingbo, Li, Xiaojie, Hu, Hongyin, Shrestha, Nawal, Li, Minjie, and Yang, Yongzhi

Subjects
FOSSIL pollen, GENOMICS, COMPARATIVE genomics, MYCORRHIZAL plants, CLIMATE change, GENE families, GENOMES
Abstract

Background Bretschneidera sinensis is an endangered relic tree species in the Akaniaceae family and is sporadically distributed in eastern Asia. As opposed to its current narrow and rare distribution, the fossil pollen of B. sinensis has been found to be frequent and widespread in the Northern Hemisphere during the Late Miocene. B. sinensis is also a typical mycorrhizal plant, and its annual seedlings exhibit high mortality rates in absence of mycorrhizal development. The chromosome-level high-quality genome of B. sinensis will help us to more deeply understand the survival and demographic histories of this relic species. Results A total of 25.39 Gb HiFi reads and 109.17 Gb Hi-C reads were used to construct the chromosome-level genome of B. sinensis , which is 1.21 Gb in length with the contig N50 of 64.13 Mb and chromosome N50 of 146.54 Mb. The identified transposable elements account for 55.21% of the genome. A total of 45,839 protein-coding genes were predicted in B. sinensis. A lineage-specific whole-genome duplication was detected, and 7,283 lineage-specific expanded gene families with functions related to the specialized endotrophic mycorrhizal adaptation were identified. The historical effective population size (N e ) of B. sinensis was found to oscillate greatly in response to Quaternary climatic changes. The N e of B. sinensis has decreased rapidly in the recent past, making its extant N e extremely lower. Our additional evolutionary genomic analyses suggested that the developed mycorrhizal adaption might have been repeatedly disrupted by environmental changes caused by Quaternary climatic oscillations. The environmental changes and an already decreased population size during the Holocene may have led to the current rarity of B. sinensis. Conclusion This is a detailed report of the genome sequences for the family Akaniaceae distributed in evergreen forests in eastern Asia. Such a high-quality genomic resource may provide critical clues for comparative genomics studies of this family in the future. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

28. The genome sequence provides insights into salt tolerance of Achnatherum splendens (Gramineae), a constructive species of alkaline grassland.

Author

Ren, Guangpeng, Jiang, Yanyou, Li, Ao, Yin, Mou, Li, Minjie, Mu, Wenjie, Wu, Ying, and Liu, Jianquan

Subjects
GRASSES, GENOME size, BRACHYPODIUM, GRASSLANDS, SPECIES, GENE families
Abstract

Summary: Achnatherum splendens Trin. (Gramineae) is a constructive species of the arid grassland ecosystem in Northwest China and is a major forage grass. It has good tolerance of salt and drought stress in alkaline habitats. Here, we report its chromosome‐level genome, determined through a combination of Illumina HiSeq sequencing, PacBio sequencing and Hi‐C technology. The final assembly of the ~1.17 Gb genome sequence had a super‐scaffold N50 of 40.3 Mb. A total of 57 374 protein‐coding genes were annotated, of which 54 426 (94.5%) genes have functional protein annotations. Approximately 735 Mb (62.37%) of the assembly were identified as repetitive elements, and among these, LTRs (40.53%) constitute the highest proportion, having made a major contribution to the expansion of genome size in A. splendens. Phylogenetic analysis revealed that A. splendens diverged from the Brachypodium distachyon–Hordeum vulgare–Aegilops tauschii subclade around 37 million years ago (Ma) and that a clade comprising these four species diverged from the Phyllostachys edulis clade ~47 Ma. Genomic synteny indicates that A. splendens underwent an additional species‐specific whole‐genome duplication (WGD) 18–20 Ma, which further promoted an increase in copies of numerous saline–alkali‐related gene families in the A. splendens genome. By transcriptomic analysis, we further found that many of these duplicated genes from this extra WGD exhibited distinct functional divergence in response to salt stress. This WGD, therefore, contributed to the strong resistance to salt stress and widespread arid adaptation of A. splendens. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

29. A multi-emitter ultrasonic flatness detecting device

Author

Yan Zhen, Shen Lingxiao, Mu Wenjie, and Zhao Gang

Subjects
010302 applied physics, Microscope, Materials science, Acoustics and Ultrasonics, business.industry, Flatness (systems theory), Acoustics, Ultrasound, Ranging, 01 natural sciences, Light scattering, law.invention, law, 0103 physical sciences, Ultrasonic sensor, business, 010301 acoustics, Common emitter, Jitter
Abstract

During the manufacture and processing of products, flatness is an important indicator of product quality. At present, flatness detection methods can be classified into contact and non-contact methods. Contact detection requires mutual contact with the test object, so the detection efficiency is low and it is easy to cause damage to the product. In the non-contact detection method, the accuracy of the microscope detection is very high, but the detection area is small and the cost is very high. Visual detection based on light scattering has high requirements for jitter and light, and strong jitter or light directly affects the detection results. The traditional ultrasonic ranging sensor requires the cooperation of the transmitting end and the receiving end, and has higher requirements on the ambient temperature. We found that when an ultrasonic array was used to generate strong ultrasound, the sound waves are reflected on the surface of the object and act on the polystyrene beads. We succeeded in detecting the flatness by comparing the speed difference of polystyrene ball at different positions on the surface of the product to be checked. This method may be applied to places where non-contact detection is required.

Published
2020

Catalog

Books, media, physical & digital resources
See catalog results