检索结果-内蒙古大学图书馆

International Conference on Computer Vision (ICCV)

作者： Chen Zhang Ganzhangqin Yuan Wenbing Tao National Key Laboratory of Science and Technology on Multi-Spectral Information Processing School of Artifical Intelligence and Automation Huazhong University of Science and Technology China TuKe Research

Recently, there has been a growing interest in learningbased explicit methods due to their ability to respect the original input and preserve details. However, the connectivity on complex structures is still difficult to infer due to the limited local shape perception, resulting in artifacts and non-watertight triangles. In this paper, we present a novel learning-based method with Delaunay triangulation to achieve high-precision reconstruction. We model the Delaunay triangulation as a dual graph, extract local geometric information from the points, and embed it into the structural representation of Delaunay triangulation in an organic way, benefiting fine-grained details reconstruction. To encourage neighborhood information interaction of edges and nodes in the graph, we introduce a local graph iteration algorithm, which is a variant of graph neural network. Moreover, a geometric constraint loss further improves the classification of tetrahedrons. Benefiting from our fully local network, a scaling strategy is designed to enable large-scale reconstruction. Experiments show that our method yields watertight and high-quality meshes. Especially for some thin structures and sharp edges, our method shows better performance than the current state-of-the-art methods. Furthermore, it has a strong adaptability to point clouds of different densities.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Deep Seam Prediction for Image Stitching Based on Selection Consistency Loss

arXiv

引用

arXiv 2023年

作者： Cheng, Senmao Yang, Fan Chen, Zhi Yuan, Nanjun Tao, Wenbing National Key Laboratory of Science and Technology on Multi-spectral Information Processing School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China

Image stitching is to construct panoramic images with wider field of vision (FOV) from some images captured from different viewing positions. To solve the problem of fusion ghosting in the stitched image, seam-driven methods avoid the misalignment area to fuse images by predicting the best seam. Currently, as standard tools of the OpenCV library, dynamic programming (DP) and GraphCut (GC) are still the only commonly used seam prediction methods despite the fact that they were both proposed two decades ago. However, GC can get excellent seam quality but poor real-time performance while DP method has good efficiency but poor seam quality. In this paper, we propose a deep learning based seam prediction method (DSeam) for the sake of high seam quality with high efficiency. To overcome the difficulty of the seam description in network and no GroundTruth for training we design a selective consistency loss combining the seam shape constraint and seam quality constraint to supervise the network learning. By the constraint of the selection of consistency loss, we implicitly defined the mask boundaries as seams and transform seam prediction into mask prediction. To our knowledge, the proposed DSeam is the first deep learning based seam prediction method for image stitching. Extensive experimental results well demonstrate the superior performance of our proposed Dseam method which is 15 times faster than the classic GC seam prediction method in OpenCV 2.4.9 with similar seam quality. Copyright © 2023, The Authors. All rights reserved.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

PSS: Point Semantic Saliency for 3D Object Detection 1st

PSS: Point Semantic Saliency for 3D Object Detection

引用

1st CAAI International Conference on Artificial Intelligence, CICAI 2021

作者： Cen, Jiajing An, Pei Chen, Gaojie Liang, Junxiong Ma, Jie National Key Laboratory of Science and Technology on Multi-spectral Information Processing School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan China

ISBN: (纸本)9783030930455

Efficient fusion on LiDAR-camera data for 3D object detection is a challenging task. Although RGB image provides sufficient texture and semantic features, some of them are unrelated to the targeted objects, which are useless and even misleading for detection task. In this paper, point semantic saliency (PSS) is proposed for precise fusion. In this scheme, physical receptive field (PRF) constraint is built to establish the relation of 2D and 3D receptive fields from camera projection model. To increase the saliency of the pixel from targeted object, we propose PSS to extract salient point feature with the guidance of RGB and semantic segmentation images, which provides 2D supplementary information for 3D detection. Comparison results and ablation studies demonstrate that PSS improves the detection performance in both localization and classification. Among the current single stage detectors, our method improves APs by 0.62 % for hard level and mAP by 0.31 %. © 2021, Springer Nature Switzerland AG.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

VAS: An Unsupervised Visible Area Sensing Module for Lane Classification

VAS: An Unsupervised Visible Area Sensing Module for Lane Cl...

引用

Big Data, information and Computer Network (BDICN), International Conference on

作者： Xing Hong Ruixiang Zhao Wu Lv Jie Ma National Key Laboratory of Science and Technology on Multi-spectral Information Processing School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan China CSSC Navigation Technology Co. Ltd Beijing China

ISBN: (数字)9798350363821

ISBN: (纸本)9798350374612

Road lane segmentation and classification play a crucial role in autonomous vehicles. Lane segmentation helps localize the position of the lane boundaries and the category information provides assistance for making decisions such as lane changing. With the development of deep learning, lane segmentation becomes faster and stronger, while classification has not received enough attention. Existing methods separate them into two networks, and can not handle complex scenes such as occlusion well. In this paper, we propose a novel network architecture combining the two tasks. To improve the robustness of classification to the scene, we also design a unsupervised visible area sensing module inspired by the simple lane geometry structure. The model was evaluated on several datasets with lane category information and get competitive results.

关键词： Geometry Deep learning Roads Network architecture Robustness Real-time systems Sensors

来源：评论

学校读者我要写书评

暂无评论

HLA: Harmonized Label Assigner for Two-stage Oriented Object Detection

HLA: Harmonized Label Assigner for Two-stage Oriented Object...

引用

Automation, Robotics and Computer Engineering (ICARCE), International Conference on

作者： Qimeng Chen Tong Zheng Liu Liu Longji Yu Zhong Chen National Key Laboratory of Science and Technology on Multi-Spectral Information Processing Huazhong University of Science and Technology Wuhan China

ISBN: (纸本)9781665475495

The existing state-of-the-arts two-stage oriented object detectors have no significant improvement in the label assignment strategies, and the most widely-used one is the so-called Max IoU Assigner (MIA). In this paper, we first illustrate that MIA may cause matching conflicts in some cases, hinder the matching of ground-truth (GT) boxes with high-quality samples, which is extremely harmful to the training process. After that, we propose a Harmonized Label Assigner (HLA) for the oriented RPN, which can automatically harmonize the assignment priority of each GT box according to the corresponding number of candidate samples, solve the matching conflicts, and improve the detection accuracy of the two-stage oriented detectors. Finally, we implement the proposed HLA on Oriented R-CNN and conduct sufficient experiments on two public datasets (MAR20 and HRSC2016). Without tricks, our HLA significantly improves the detection accuracy of the detector to 83.97% mAP (on MAR20) and 90.42% mAP (on HRSC2016), respectively.

关键词： Training Automation Detectors Object detection Robots

来源：评论

学校读者我要写书评

暂无评论

Semantic Segmentation Algorithm of Remote Sensing Images Based on Improved Panoptic

Semantic Segmentation Algorithm of Remote Sensing Images Bas...

引用

Automation, Robotics and Computer Engineering (ICARCE), International Conference on

作者： Tong Zheng Liu Liu Qimeng Chen Zhong Chen Longji Yu Zeyu Zhao National Key Laboratory of Science and Technology on Multi-Spectral Information Processing Huazhong University of Science and Technology Wuhan China

ISBN: (纸本)9781665475495

Semantic segmentation of images can play an important role in land use, building extraction, road extraction and vehicle detection using remote sensing images. In the scenario of applying Panoptic FPN algorithm for remote sensing images, we found that the decoder of this algorithm is not robust in feature extraction and cannot do weighted enhancement of key spaces and channels, and its encoder loses a lot of high-dimensional semantic features when simply fusing semantic information at all levels by simple addition. To address these two problems, we propose the Se-Resnext encoder based on the attention mechanism and the Concat-based feature fusion mechanism, respectively, and verify the effectiveness of the methods through experiments. The accuracy of semantic segmentation is improved in both suichang dataset and posdam dataset.

关键词： Training Fuses Semantic segmentation Vehicle detection Roads Semantics Feature extraction

来源：评论

学校读者我要写书评

暂无评论

PCA-W: An Simulation Algorithm for Waterlogging in Ground Point Cloud Based On Principal Component Analysis

PCA-W: An Simulation Algorithm for Waterlogging in Ground Po...

引用

Big Data, information and Computer Network (BDICN), International Conference on

作者： Yucong Duan Yifan Zhang Ziwen Xu Jie Ma National Key Laboratory of Science and Technology on Multi-spectral Information Processing School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan Hubei China China Ship Development and Design Center Wuhan Hubei China

ISBN: (数字)9798350363821

ISBN: (纸本)9798350374612

In the context of autonomous driving, ground point clouds play an important role in system decision-making. However, in the case of adverse weather, ground point clouds may suffer from a certain degree of missing data, which will have a negative impact on some downstream tasks of autonomous driving. With the development of autonomous driving algorithms, deep learning has achieved better results in this field. However, due to the lack of point cloud data sets under adverse weather conditions, the network cannot be effectively trained. This paper proposes a simulation algorithm for ground point clouds with water accumulation, which simulates the phenomenon of missing data in the ground point clouds due to road water accumulation in adverse weather. As a data augmentation method, it can expand the data set for downstream tasks. We evaluated the method on the STF dataset and proved its effectiveness.

关键词： Point cloud compression Deep learning Meteorological radar Laser radar Roads Data augmentation Task analysis

来源：评论

学校读者我要写书评

暂无评论

De-DETR: A More Accurate Real-Time Defect Detection Method

De-DETR: A More Accurate Real-Time Defect Detection Method

引用

Big Data, information and Computer Network (BDICN), International Conference on

作者： Zhongmu Yue Hao Xu Yifan Zhang Ziwen Xu Jie Ma National Key Laboratory of Science and Technology on Multi-spectral Information Processing School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan Hubei China China Ship Development and Design Center Wuhan Hubei China

ISBN: (数字)9798350363821

ISBN: (纸本)9798350374612

In industrial production, defect detection technology plays a very important role in improving product yield and increasing enterprise efficiency. At present, defect detection technology based on deep learning mainly focuses on convolutional neural networks. With the gradual application of the Transformer architecture represented by the DETR[1] model in the image field, there is a new direction for the update of defect detection technology. This paper proposes a defect detection method based on DETR. This method focuses on improving the encoding method of defect information, further improving the information fusion efficiency of deep features and shallow features, while maximizing the retention of deep semantic information, effectively improving the overall performance of the model. Experiments show that this method is superior to excellent real-time detection models such as YOLO and Rt-DETR[2] in terms of defect detection effects on the NEU-DET dataset.

关键词： YOLO Location awareness Image coding Image recognition Target recognition Semantics Production

来源：评论

学校读者我要写书评

暂无评论

Hidden Object Detection Based on Probabilistic Fuzzy Fusion and Fisher Vectors in Passive Millimeter-Wave Images

Hidden Object Detection Based on Probabilistic Fuzzy Fusion ...

引用

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Bo Fang Dong Zhu Yayun Cheng Fei Hu School of Electronic Information and Communications Huazhong University of Science and Technology Wuhan China National key Laboratory of Science and Technology on Multi-Spectral Information Processing Wuhan China School of Electronics and Information Engineering Harbin Institute of Technology Harbin China

Millimeter-wave radiometric imaging technology holds enormous potential for hidden-object detection in the application of security inspection. Existing hidden-object detection methods based on pixel or regional processing have limited detection performance on the low signal-to-clutter ratio or the targets with multiple sizes or irregular shapes. This paper proposes a new detection method combining the probabilistic fuzzy fusion and Fisher vectors processing for hidden object detection in multiple polarization passive millimeter-wave images. Experimental results employing the real data collected from multiple Polarization Scanning Imaging Radiometer demonstrate that the proposed method improves the detection performance and the extraction accuracy of object contours compared with the state-of-the-art methods.

关键词：

来源：评论

学校读者我要写书评

暂无评论

HS-FPN: High Frequency and Spatial Perception FPN for Tiny Object Detection 39

HS-FPN: High Frequency and Spatial Perception FPN for Tiny O...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Shi, Zican Hu, Jing Ren, Jie Ye, Hengkang Yuan, Xuyang Ouyang, Yan He, Jia Ji, Bo Guo, Junyu Huazhong University of Science and Technology Wuhan China National Key Laboratory of Science and Technology on Multi-Spectral Information Processing Wuhan China Air Force Early Warning Academy Wuhan China Chinese People’s Liberation Army 95841 troops China

ISBN: (纸本)157735897X

The introduction of Feature Pyramid Network (FPN) has significantly improved object detection performance. However, substantial challenges remain in detecting tiny objects, as their features occupy only a very small proportion of the feature maps. Although FPN integrates multi-scale features, it does not directly enhance or enrich the features of tiny objects. Furthermore, FPN lacks spatial perception ability. To address these issues, we propose a novel High Frequency and Spatial Perception Feature Pyramid Network (HS-FPN) with two innovative modules. First, we designed a high frequency perception module (HFP) that generates high frequency responses through high pass filters. These high frequency responses are used as mask weights from both spatial and channel perspectives to enrich and highlight the features of tiny objects in the original feature maps. Second, we developed a spatial dependency perception module (SDP) to capture the spatial dependencies that FPN lacks. Our experiments demonstrate that detectors based on HS-FPN exhibit competitive advantages over state-of-the-art models on the AI-TOD dataset for tiny object detection. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：