請用此 Handle URI 來引用此文件：
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/81163完整後設資料紀錄
| DC 欄位 | 值 | 語言 |
|---|---|---|
| dc.contributor.advisor | 徐宏民(Winston H. Hsu) | |
| dc.contributor.author | Yu-Kai Huang | en |
| dc.contributor.author | 黃郁凱 | zh_TW |
| dc.date.accessioned | 2022-11-24T03:33:46Z | - |
| dc.date.available | 2021-08-11 | |
| dc.date.available | 2022-11-24T03:33:46Z | - |
| dc.date.copyright | 2021-08-11 | |
| dc.date.issued | 2021 | |
| dc.date.submitted | 2021-08-06 | |
| dc.identifier.citation | [1]Radhakrishna Achanta, Appu Shaji, Kevin Smith, Aurelien Lucchi, Pascal Fua, andSabine Süsstrunk. Slic superpixels compared to stateoftheart superpixel methods.IEEE transactions on pattern analysis and machine intelligence,34(11):2274–2282,2012. [2]Talha Ahmad Siddiqui, Rishi Madhok, and Matthew O’Toole. An extensible multisensor fusion framework for 3d imaging. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pages 1008–1009,2020. [3]Cesar Cadena, Anthony R Dick, and Ian D Reid. Multimodal autoencoders asjointestimatorsforroboticssceneunderstanding. InRobotics: Science and Systems,volume 5, page 1, 2016. [4]Holger Caesar, Varun Bankiti, Alex H. Lang, Sourabh Vora, Venice Erin Liong, Qiang Xu, Anush Krishnan, Yu Pan, Giancarlo Baldan, and Oscar Beijbom. nuscenes: A multimodal dataset for autonomous driving.arXiv preprintarXiv:1903.11027, 2019. [5]SimonChadwick, WillMaddetn, andPaulNewman. Distantvehicledetectionusingradar and vision. In2019 International Conference on Robotics and Automation(ICRA), pages 8311–8317, 2019. [6]JiaRen Chang and YongSheng Chen. Pyramid stereo matching network. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition,pages 5410–5418, 2018. [7]Yun Chen, Bin Yang, Ming Liang, and Raquel Urtasun. Learning joint 2d3d representations for depth completion. InProceedings of the IEEE International Conference on Computer Vision, pages 10023–10032, 2019. [8]Xinjing Cheng, Peng Wang, and Ruigang Yang. Learning depth with convolutionalspatial propagation network.IEEE transactions on pattern analysis and machineintelligence, 2019. [9]Xuelian Cheng, Yiran Zhong, Yuchao Dai, Pan Ji, and Hongdong Li. Noiseawareunsupervised deep lidarstereo fusion. InProceedings of the IEEE Conference onComputer Vision and Pattern Recognition, pages 6339–6348, 2019. [10]Liam Daniel, Andrew Stove, Edward Hoare, Dominic Phippen, Mike Cherniakov,Bernie Mulgrew, and Marina Gashinova. Application of doppler beam sharpeningfor azimuth refinement in prospective lowthz automotive radars.IET Radar, Sonar Navigation, 12(10):1121–1130, 2018. [11]David Eigen, Christian Puhrsch, and Rob Fergus. Depth map prediction from asingle image using a multiscale deep network. InAdvances in neural informationprocessing systems, pages 2366–2374, 2014. [12]Andreas Geiger, Philip Lenz, Christoph Stiller, and Raquel Urtasun. Vision meetsrobotics: The kitti dataset.The International Journal of Robotics Research,32(11):1231–1237, 2013. [13]Andreas Geiger, Philip Lenz, and Raquel Urtasun. Are we ready for autonomousdriving? the kitti vision benchmark suite. InConference on Computer Vision and Pattern Recognition (CVPR), 2012. [14]Andreas Geiger, Julius Ziegler, and Christoph Stiller. Stereoscan: Dense 3d reconstruction in realtime. In2011 IEEE intelligent vehicles symposium (IV), pages963–968, 2011. [15]Simon Hawe, Martin Kleinsteuber, and Klaus Diepold. Dense disparity maps fromsparse disparity measurements. In2011 International Conference on Computer Vision, pages 2126–2133, 2011. [16]HeikoHirschmuller. Accurateandefficientstereoprocessingbysemiglobalmatching and mutual information. In2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05), volume 2, pages 807–814, 2005. [17]YuKaiHuang,TsungHanWu,YuehChengLiu,andWinstonH.Hsu. Indoordepthcompletion with boundary consistency and selfattention. InProceedings of theIEEE/CVF International Conference on Computer Vision (ICCV) Workshops, Oct2019. [18]Alex Kendall, Hayk Martirosyan, Saumitro Dasgupta, Peter Henry, Ryan Kennedy,Abraham Bachrach, and Adam Bry. Endtoend learning of geometry and contextfor deep stereo regression. InProceedings of the IEEE International Conference onComputer Vision, pages 66–75, 2017. [19]Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization.arXiv preprint arXiv:1412.6980, 2014. [20]JaeHan Lee and ChangSu Kim. Monocular depth estimation using relative depthmaps. InProceedings of the IEEE Conference on Computer Vision and PatternRecognition, pages 9729–9738, 2019. [21]Chi Li and Zhiguo Cao. Lidarstereo: Dense depth estimation from sparse lidarand stereo images. InProceedings of the 2020 5th International Conference onMultimedia Systems and Signal Processing, pages 11–15, 2020. [22]LeeKang Liu, Stanley H Chan, and Truong Q Nguyen. Depth reconstruction fromsparse samples: Representation, algorithm, and sampling.IEEE Transactions onImage Processing, 24(6):1983–1996, 2015. [23]Fangchang Ma, Guilherme Venturelli Cavalheiro, and Sertac Karaman. Selfsupervised sparsetodense: Selfsupervised depth completion from lidar andmonocular camera. In2019 International Conference on Robotics and Automation(ICRA), pages 3288–3295, 2019. [24]FangchangMalandSertacKaraman. Sparsetodense: Depthpredictionfromsparsedepth samples and a single image. In2018 IEEE International Conference onRobotics and Automation (ICRA), pages 1–8, 2018. [25]Suresh B Marapane and Mohan M Trivedi. Regionbased stereo analysis for roboticapplications.IEEE Transactions on Systems, Man, and Cybernetics, 19(6):1447–1464, 1989. [26]Nikolaus Mayer, Eddy Ilg, Philip Hausser, Philipp Fischer, Daniel Cremers, AlexeyDosovitskiy, and Thomas Brox. A large dataset to train convolutional networksfor disparity, optical flow, and scene flow estimation. InProceedings of the IEEEConference on Computer Vision and Pattern Recognition, pages 4040–4048, 2016. [27]Moritz Menze and Andreas Geiger. Object scene flow for autonomous vehicles. InConference on Computer Vision and Pattern Recognition (CVPR), 2015. [28]Lazaros Nalpantidis and Antonios Gasteratos. Stereo vision for robotic applications in the presence of nonideal lighting conditions.Image and Vision Computing, 28(6):940–951, 2010. [29]Felix Nobis, Maximilian Geisslinger, Markus Weber, Johannes Betz, and MarkusLienkamp. Adeeplearningbasedradarandcamerasensorfusionarchitectureforobject detection. In2019 Sensor Data Fusion: Trends, Solutions, Applications (SDF),pages 1–7, 2019. [30]Kihong Park, Seungryong Kim, and Kwanghoon Sohn. Highprecision depth estimation with the 3d lidar and stereo fusion. In2018 IEEE International Conferenceon Robotics and Automation (ICRA), pages 2156–2163, 2018. [31]Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, GregoryChanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. Pytorch: An imperative style, highperformance deep learning library. InAdvances inNeural Information Processing Systems, pages 8024–8035, 2019. [32]Matteo Poggi, Davide Pallotti, Fabio Tosi, and Stefano Mattoccia. Guided stereomatching. InProceedings of the IEEE Conference on Computer Vision and PatternRecognition, pages 979–988, 2019. [33]Matteo Poggi, Fabio Tosi, and Stefano Mattoccia. Quantitative evaluation of confidence measures in a machine learning world. InProceedings of the IEEE International Conference on Computer Vision, pages 5228–5237, 2017. [34]Jiaxiong Qiu, Zhaopeng Cui, Yinda Zhang, Xingdi Zhang, Shuaicheng Liu, BingZeng, and Marc Pollefeys. Deeplidar: Deep surface normal guided depth predictionforoutdoorscenefromsparselidardataandsinglecolorimage. InProceedingsoftheIEEE Conference on Computer Vision and Pattern Recognition, pages 3313–3322,2019. [35]Olaf Ronneberger, Philipp Fischer, and Thomas Brox. Unet: Convolutional networks for biomedical image segmentation. InInternational Conference on Medicalimage computing and computerassisted intervention, pages 234–241, 2015.[36]Antonio Rubio, LongLong Yu, Edgar SimoSerra, and Francesc MorenoNoguer.Bass: boundaryaware superpixel segmentation. In2016 23rd International Conference on Pattern Recognition (ICPR), pages 2824–2829, 2016. [37]Marcel Sheeny, Andrew Wallace, and Sen Wang. 300 ghz radar object recognitionbased on deep neural networks and transfer learning.IET Radar, Sonar Navigation, 14(10):1483–1493, 2020. [38]ShreyasSShivakumar,KartikMohta,BerndPfrommer,VijayKumar,andCamilloJTaylor. Real time dense depth estimation by fusing stereo with sparse depth measurements. In2019 International Conference on Robotics and Automation (ICRA),pages 6482–6488, 2019. [39]Shreyas S Shivakumar, Ty Nguyen, Ian D Miller, Steven W Chen, Vijay Kumar,and Camillo J Taylor. Dfusenet: Deep fusion of rgb and sparse depth informationfor image guided dense depth completion. In2019 IEEE Intelligent TransportationSystems Conference (ITSC), pages 13–20, 2019. [40]Aristotle Spyropoulos, Nikos Komodakis, and Philippos Mordohai. Learning to detectgroundcontrolpointsforimprovingtheaccuracyofstereomatching.InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages1621–1628, 2014. [41]JonasUhrig, NickSchneider, LukasSchneider, UweFranke, ThomasBrox, andAndreasGeiger. Sparsityinvariantcnns. In2017 international conferenceon 3D Vision(3DV), pages 11–20, 2017. [42]Michael Van den Bergh, Xavier Boix, Gemma Roig, Benjamin de Capitani, and LucVan Gool. Seeds: Superpixels extracted via energydriven sampling. InEuropeanconference on computer vision, pages 13–26, 2012. [43]TsunHsuanWang,HouNingHu,ChiehHubertLin,YiHsuanTsai,WeiChenChiu,and Min Sun. 3d lidar and stereo fusion using stereo matching network with conditional cost volume normalization. In2019 IEEE/RSJ International Conference onIntelligent Robots and Systems (IROS), pages 5895–5902, 2019. [44]YanWang, WeiLunChao, DivyanshGarg, BharathHariharan, MarkCampbell, andKilianQWeinberger. Pseudolidarfromvisualdepthestimation: Bridgingthegapin3d object detection for autonomous driving. InProceedings of the IEEE Conferenceon Computer Vision and Pattern Recognition, pages 8445–8453, 2019. [45]Adam Wolff, Shachar Praisler, Ilya Tcenov, and Guy Gilboa. Superpixel sampler:a datadriven approach for depth sampling and reconstruction. In2020 IEEE International Conference on Robotics and Automation (ICRA), pages 2588–2594, 2020. [46]Jian Yao, Marko Boben, Sanja Fidler, and Raquel Urtasun. Realtime coarsetofinetopologically preserving segmentation. InProceedings of the IEEE Conference onComputer Vision and Pattern Recognition, pages 2947–2955, 2015. [47]Yurong You, Yan Wang, WeiLun Chao, Divyansh Garg, Geoff Pleiss, Bharath Hariharan, Mark Campbell, and Kilian Q Weinberger. Pseudolidar++: Accurate depthfor 3d object detection in autonomous driving. InICLR, 2020. [48]Feihu Zhang, Victor Prisacariu, Ruigang Yang, and Philip HS Torr. Ganet: Guidedaggregation net for endtoend stereo matching. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 185–194, 2019. [49]Ke Zhang, Jiangbo Lu, and Gauthier Lafruit. Crossbased local stereo matchingusing orthogonal integral images.IEEE transactions on circuits and systems forvideo technology, 19(7):1073–1079, 2009. [50]Yinda Zhang and Thomas Funkhouser. Deep depth completion of a single rgbdimage. InProceedings of the IEEE Conference on Computer Vision and PatternRecognition, pages 175–185, 2018. [51]Zhengyou Zhang. Determining the epipolar geometry and its uncertainty: A review.International journal of computer vision, 27(2):161–195, 1998. [52]Yiqi Zhong, ChoYing Wu, Suya You, and Ulrich Neumann. Deep rgbd canonicalcorrelationanalysisforsparsedepthcompletion. InAdvances in Neural InformationProcessing Systems, pages 5331–5341, 2019. | |
| dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/81163 | - |
| dc.description.abstract | 深度預測在日常生活應用上扮演重要角色，例如機器人領域、3D重建、擴增實境及自駕車。過去許多方法使用額外的深度訊號如光達或雷達讓深度預測更加準確（引導式深度預測），但這樣的訊號覆蓋率不高，以及分佈不均勻。為了能夠完整利用這樣的深度訊號，我們根據稀疏訊號的特性，提出了Sparse Signal Superdensity (S3)方法，能夠將稀疏不均勻的訊號擴展並得到密度較高的深度圖，外加一張衡量擴展程度的信心圖。S3可以應用在各種引導式深度預測上，做到端對端訓練，並包括不同應用階段：輸入、匹配代價卷、輸出和3D空間。大量的實驗展現了我們方法在光達和雷達的有效性、魯棒性及應用上的彈性。 | zh_TW |
| dc.description.provenance | Made available in DSpace on 2022-11-24T03:33:46Z (GMT). No. of bitstreams: 1 U0001-0508202121521300.pdf: 13500008 bytes, checksum: f1928f83ca1591b472485fc07f1ecbe3 (MD5) Previous issue date: 2021 | en |
| dc.description.tableofcontents | Acknowledgement i 摘要 ii Abstract iii Contents iv List of Figures vi List of Tables vii Chapter 1 Introduction 1 Chapter 2 Related Work 5 2.1 Guided Mono Estimation 5 2.2 Guided Stereo Estimation 5 2.3 Signal Expansion 6 Chapter 3 Method 8 3.1 Intuition of Sparse Signal Superdensity 8 3.2 Learnable Sparse Signal Superdensity 9 Chapter 4 Application of S3 12 4.1 Guidance on Input and Output 13 4.2 Guidance on Cost Volume 13 4.2.1 Guided Stereo Matching (GSM) 14 4.2.2 Conditional Cost Volume Normalization (CCVNorm) 15 4.3 Guidance on 3D Space 17 Chapter 5 Experiment 19 5.1 Experimental Setting 19 5.1.1 Dataset 19 5.1.2 Training Protocol 20 5.1.3 Implementation Detail 20 5.1.4 Evaluation Metric 21 5.2 Guidance Experiment 21 5.2.1 Guidance on Input and Output 21 5.2.2 Guidance on Cost Volume 22 5.2.3 Guidance on 3D Space 23 5.3 Radar Guidance 24 5.4 Ablation Study 25 5.4.1 Effectiveness of Each Component 25 5.4.2 Sparsity Expansion 26 5.4.3 Robustness 27 Chapter 6 Conclusion 29 References 30 | |
| dc.language.iso | en | |
| dc.subject | 稀疏訊號 | zh_TW |
| dc.subject | 引導式深度預測 | zh_TW |
| dc.subject | 深度預測 | zh_TW |
| dc.subject | 光達和雷達 | zh_TW |
| dc.subject | 深度學習 | zh_TW |
| dc.subject | Guided Depth Estimation | en |
| dc.subject | Sparse Signal | en |
| dc.subject | Depth Estimation | en |
| dc.subject | Deep Learning | en |
| dc.subject | LiDAR and Radar | en |
| dc.title | 基於深度稀疏訊號之可學習式擴展架構用以深度預測 | zh_TW |
| dc.title | S3: Learnable Sparse Signal Superdensity for Guided Depth Estimation | en |
| dc.date.schoolyear | 109-2 | |
| dc.description.degree | 碩士 | |
| dc.contributor.oralexamcommittee | 余能豪(Hsin-Tsai Liu),陳文進(Chih-Yang Tseng),陳奕廷,葉梅珍 | |
| dc.subject.keyword | 深度學習,深度預測,稀疏訊號,引導式深度預測,光達和雷達, | zh_TW |
| dc.subject.keyword | Deep Learning,Depth Estimation,Sparse Signal,Guided Depth Estimation,LiDAR and Radar, | en |
| dc.relation.page | 37 | |
| dc.identifier.doi | 10.6342/NTU202102130 | |
| dc.rights.note | 同意授權(限校園內公開) | |
| dc.date.accepted | 2021-08-09 | |
| dc.contributor.author-college | 電機資訊學院 | zh_TW |
| dc.contributor.author-dept | 資訊工程學研究所 | zh_TW |
| 顯示於系所單位： | 資訊工程學系 | |
文件中的檔案：
| 檔案 | 大小 | 格式 | |
|---|---|---|---|
| U0001-0508202121521300.pdf 授權僅限NTU校內IP使用（校園外請利用VPN校外連線服務） | 13.18 MB | Adobe PDF |
系統中的文件，除了特別指名其著作權條款之外，均受到著作權保護，並且保留所有的權利。