[{"data":1,"prerenderedAt":600},["ShallowReactive",2],{"method-bundlefusion2017":3},{"method":4,"reference":60,"equipment":82,"figures":128,"results":129},{"id":5,"label":6,"shortName":7,"title":8,"year":9,"era":10,"cluster":11,"scope":12,"keyIdeaZh":13,"keyIdeaEn":14,"fulltextStatus":15,"publicationStatus":16,"recommendation":17,"constructionRelevance":18,"validationEnvironment":19,"strengths":24,"limitations":30,"sensors":36,"platform":38,"estimator":41,"association":42,"timeModel":43,"deskew":44,"loopClosure":45,"globalOptimization":46,"mapRepresentation":47,"prior":48,"outputGeometry":49,"compute":50,"codeUrl":51,"codeLicense":52,"relatedVersions":53},"bundlefusion2017","Dai et al., 2017a","BundleFusion","BundleFusion: Real-Time Globally Consistent 3D Reconstruction Using On-the-Fly Surface Reintegration",2017,"classic","C08","full_slam_with_global_correction","BundleFusion 在每一影格都考慮完整的 RGB-D 歷史資料，以分塊（chunk）的階層式區域到全域最佳化，結合稀疏 SIFT 特徵與稠密幾何、光度對應，即時求得經 BA 的全域位姿。位姿更新後，系統即時將受影響影格從 TSDF 中移除並以新位姿重新融合（re-integration），使稠密模型保持全域一致。由於每張影格都與全部歷史比對，迴圈閉合是隱式處理的。","BundleFusion solves real-time globally bundle-adjusted poses over the whole RGB-D history with a hierarchical sparse-plus-dense optimisation and de-\u002Fre-integrates frames in a hashed TSDF whenever poses change.","full_text_reviewed","peer_reviewed_published","main_body","論文未報告營建測試；以作者自錄大型室內掃描與 TUM、ICL-NUIM 等基準評估。約 14 分鐘的資料量上限、雙 GPU 需求，以及主記憶體隨影格數線性成長（Apt 0 需 20 GB），限制其直接用於大範圍工地掃描（推論）。作者也指出缺乏色彩紋理的牆面會使迴圈難以閉合，並使部分影格無法通過對應篩選而未完成配準（Sec. 7.2、Table 8），與未粉刷牆面的室內工地情境相關（推論）。",[20,21,22,23],"public_benchmark","simulation","controlled_experiment","independent_reference",[25,26,27,28,29],"Global consistency with quality described as on par with offline methods (abstract)","Robust tracking with relocalization from gross failures (abstract; Fig. 6)","Lowest ATE RMSE on all four ICL-NUIM living-room sequences (kt1 tied with VoxelHashing at 0.4 cm) and on three of four augmented ICL-NUIM sequences (Tables 3 and 7)","Lowest surface error on ICL-NUIM, 0.5 to 0.8 cm mean distance (Table 6)","Loop-closure precision reaches 100% after optimisation pruning at 39 to 48% recall on augmented ICL-NUIM (Table 2)",[31,32,33,34,35],"Sparse keypoint mismatches of a few pixels and depth noise can propagate into local misalignments (limitations)","The current implementation runs on two GPUs (Titan X and Titan Black in the performance evaluation), and the current hardware configuration limits scans to about 25,000 RGB-D frames, roughly 14 minutes at 30 Hz (limitations; Fig. 4)","On augmented ICL-NUIM Office 1 the method has difficulty closing the loop because part of the trajectory covers a wall with little to no colour features (ATE RMSE 15.3 cm) (Sec. 7.2; Table 7)","Frames on untextured walls or with occluded sensor fail the correspondence filters and stay unregistered (Table 8)","Parameters must be relaxed for noisier sensors such as Kinect (Sec. 6 Parameters)",[37],"RGB-D",[39,40],"hand-held Structure Sensor mounted to an iPad Air, streaming compressed RGB-D over a wireless network to a desktop that runs the optimisation and streams visual feedback back to the iPad","public datasets: TUM RGB-D (hand-held Kinect), SUN3D (Asus Xtion), NYU2 (Kinect), synthetic ICL-NUIM and augmented ICL-NUIM","Two-level hierarchical pose-only optimisation: intra-chunk alignment of 11 consecutive frames (1-frame overlap) and inter-chunk alignment of chunk keyframes with aggregated feature sets; energy = sparse SIFT correspondence distances plus dense photometric (luminance gradient) and point-to-plane geometric terms on 80x60 downsampled frames with linearly increasing dense weight; Gauss-Newton with a GPU data-parallel PCG solver (Jacobi preconditioner), warm-started from the previous frame; about 20 times faster than Ceres on a 101-keyframe sparse problem","GPU SIFT matched against all previous frames (about 150 features per frame, 250 per keyframe) and filtered by a Kabsch-based key point correspondence filter (max residual 0.02 m, condition number limit 100), a surface-area filter (0.032 m2) and two-sided dense geometric and photometric verification (Nmin = 5); after each optimisation all correspondences of a frame pair with residual above 0.05 m are pruned; dense terms use frame pairs within 60 degrees and nonzero overlap","discrete poses (all frames, grouped in chunks)","not_reported","implicit: each frame is globally correlated to all previous frames, so no explicit loop detection; relocalization after gross tracking failure (abstract; introduction)","real-time global pose optimisation with on-the-fly TSDF de-integration and re-integration when poses change; 3D keypoint positions stay fixed (only camera poses are optimised), and the dense term at the global keyframe level is run only after the user ends scanning, so online global alignment relies on sparse correspondences with dense terms inside chunks","TSDF in sparse voxel hashing with 8x8x8 voxel blocks, default 4 mm voxels (1 cm also evaluated); every frame stored with integrated and optimised poses and the 10 frames with the largest pose change re-integrated per new frame","none","textured surface mesh from the TSDF (authors contrast it with the point-cloud output of ElasticFusion) plus globally optimised per-frame poses","Intel Core i7 3.4 GHz CPU with 32 GB RAM; NVIDIA GeForce GTX Titan X for volumetric reconstruction and GTX Titan Black for correspondence search and pose optimisation; CUDA 7.0; frame rate well beyond 30 Hz; CPU RAM grows linearly with sequence length (20 GB for Apt 0, 34.7 GB for a 14,785-frame SUN3D scan) while GPU memory is 1.9 GB (1 cm) or 5.3 GB (4 mm) for Apt 0","https:\u002F\u002Fgithub.com\u002Fniessner\u002FBundleFusion","CC BY-NC-SA 4.0 (LICENSE.txt)",[54,58],{"relation":55,"title":56,"doi_or_url":57},"preprint","BundleFusion: Real-time Globally Consistent 3D Reconstruction using On-the-fly Surface Re-integration (arXiv v1-v3)","https:\u002F\u002Farxiv.org\u002Fabs\u002F1604.01093",{"relation":59,"title":7,"doi_or_url":51},"code_release",{"id":5,"kind":61,"shortName":7,"title":8,"authors":62,"year":9,"venue":68,"venueType":69,"publisher":70,"volumeIssuePages":71,"doi":72,"arxivId":73,"url":57,"firstPublicDate":74,"publicationStatus":16,"metadataStatus":75,"fulltextStatus":15,"era":10,"classicReason":76,"codeUrl":51,"cluster":11,"topics":77,"mdpi":78,"verification":79,"label":6,"fulltextRoute":80,"versionRead":81,"addedByCensus":78},"method",[63,64,65,66,67],"Angela Dai","Matthias Nießner","Michael Zollhöfer","Shahram Izadi","Christian Theobalt","ACM Transactions on Graphics","journal","ACM","36(3):1-18 (Crossref pages; article number not_reported)","10.1145\u002F3054739","1604.01093","2016-04-05","metadata_verified","necessary technical node (first public 2016-04-05 on arXiv): online global bundle-adjusted pose optimisation with TSDF de-integration and re-integration, removing explicit loop-closure detection.",[11],false,"corrected","arXiv","arXiv 1604.01093v3 (2017-02-07, 19 pages, preprint formatted for ACM TOG with placeholder DOI); ACM TOG 36(3) Article 24 version of record not compared",[83,90,96,103,107,110,116,121,125],{"category":84,"model":85,"canonical":85,"role":86,"dataset":87,"specs":88,"locator":89},"rgbd","Structure Sensor","method input","BundleFusion captured sequences (Apt 0 to 2, Copyroom, Office 0 to 3)","RGB-D stream at 30 Hz with 640x480 colour and depth; another colour resolution of 1296x968 mentioned for SIFT timing; footnote links structure.io and the acknowledgements thank Occipital for hardware donations, but the manufacturer is not named","Sec. 6; Sec. 7.3; Acknowledgments",{"category":91,"model":92,"canonical":92,"role":86,"dataset":93,"specs":94,"locator":95},"platform","iPad Air","BundleFusion captured sequences","Structure Sensor mounted on it; receives live visual feedback; data sent over wireless network with zlib depth and jpeg colour compression","Sec. 6",{"category":97,"model":98,"canonical":98,"role":99,"dataset":100,"specs":101,"locator":102},"compute","Intel Core i7 3.4GHz CPU (32GB RAM)","compute for runtime",null,"desktop host","Sec. 6 Performance and Convergence",{"category":97,"model":104,"canonical":104,"role":99,"dataset":100,"specs":105,"locator":106},"NVIDIA GeForce GTX Titan X","12 GB; volumetric reconstruction","Sec. 6; Fig. 4",{"category":97,"model":108,"canonical":108,"role":99,"dataset":100,"specs":109,"locator":106},"GTX Titan Black","correspondence search and global pose optimisation",{"category":84,"model":111,"canonical":111,"role":112,"dataset":113,"specs":114,"locator":115},"Kinect","dataset sensor","TUM RGB-D","hand-held Kinect sequences","Sec. 6 Quantitative Comparison",{"category":117,"model":118,"canonical":118,"role":119,"dataset":113,"specs":120,"locator":115},"other","calibrated motion capture system","reference or ground truth","ground-truth camera poses of TUM RGB-D",{"category":84,"model":122,"canonical":122,"role":112,"dataset":123,"specs":44,"locator":124},"Asus Xtion sensor","SUN3D","Sec. 7.1",{"category":84,"model":111,"canonical":111,"role":112,"dataset":126,"specs":127,"locator":124},"NYU2","all 464 scenes reconstructed qualitatively",[],{"totalRows":130,"groupCount":131,"groups":132,"others":537},85,15,[133,263,372,484],{"slug":134,"group":135,"sourceId":5,"sourceLabel":6,"table":136,"selfRows":137,"metrics":138,"seqs":143,"entrants":154,"cells":179,"outcomes":257,"locators":258,"hardware":259,"wordings":260,"notes":261},"bundlefusion2017-table-3","bundlefusion2017:Table 3","Table 3",12,[139],{"label":140,"unit":141,"statistic":142,"alignment":44},"ATE RMSE","cm","RMSE",[144,148,150,152],{"dataset":145,"sequence":146,"environment":147},"ICL-NUIM","kt0","synthetic living room",{"dataset":145,"sequence":149,"environment":147},"kt1",{"dataset":145,"sequence":151,"environment":147},"kt2",{"dataset":145,"sequence":153,"environment":147},"kt3",[155,157,159,161,165,168,171,173,175,177],{"name":156,"methodId":100,"linkable":78,"proposed":78,"self":78},"DVO SLAM",{"name":158,"methodId":100,"linkable":78,"proposed":78,"self":78},"RGB-D SLAM",{"name":160,"methodId":100,"linkable":78,"proposed":78,"self":78},"MRSMap",{"name":162,"methodId":163,"linkable":164,"proposed":78,"self":78},"Kintinuous","kintinuous2015",true,{"name":166,"methodId":167,"linkable":164,"proposed":78,"self":78},"VoxelHashing","voxelhashing2013",{"name":169,"methodId":170,"linkable":164,"proposed":78,"self":78},"Elastic Fusion","elasticfusion2015",{"name":172,"methodId":100,"linkable":78,"proposed":78,"self":78},"Redwood (rigid)",{"name":174,"methodId":5,"linkable":164,"proposed":78,"self":164},"BundleFusion ablation: Ours (s), sparse only",{"name":176,"methodId":5,"linkable":164,"proposed":78,"self":164},"BundleFusion ablation: Ours (sd), sparse and local dense",{"name":178,"methodId":5,"linkable":164,"proposed":164,"self":164},"BundleFusion (Ours)",[180,184,187,190,193,195,197,199,201,203,205,207,209,211,213,214,216,219,221,222,223,226,227,228,230,233,234,236,238,240,242,244,245,247,248,250,251,254,255,256],[181,181,181,182,183,181,183,183,181],0,10.4,-1,[181,181,185,186,183,181,183,183,181],1,2.9,[181,181,188,189,183,181,183,183,181],2,19.1,[181,181,191,192,183,181,183,183,181],3,15.2,[185,181,181,194,183,181,183,183,181],2.6,[185,181,185,196,183,181,183,183,181],0.8,[185,181,188,198,183,181,183,183,181],1.8,[185,181,191,200,183,181,183,183,181],43.3,[188,181,181,202,183,181,183,183,181],20.4,[188,181,185,204,183,181,183,183,181],22.8,[188,181,188,206,183,181,183,183,181],18.9,[188,181,191,208,183,181,183,183,181],109,[191,181,181,210,183,181,183,183,181],7.2,[191,181,185,212,183,181,183,183,181],0.5,[191,181,188,185,183,181,183,183,181],[191,181,191,215,183,181,183,183,181],35.5,[217,181,181,218,183,181,183,183,181],4,1.4,[217,181,185,220,183,181,183,183,181],0.4,[217,181,188,198,183,181,183,183,181],[217,181,191,137,183,181,183,183,181],[224,181,181,225,183,181,183,183,181],5,0.9,[224,181,185,225,183,181,183,183,181],[224,181,188,218,183,181,183,183,181],[224,181,191,229,183,181,183,183,181],10.6,[231,181,181,232,183,181,183,183,181],6,25.6,[231,181,185,191,183,181,183,183,181],[231,181,188,235,183,181,183,183,181],3.3,[231,181,191,237,183,181,183,183,181],6.1,[239,181,181,225,183,181,183,183,181],7,[239,181,185,241,183,181,183,183,181],1.2,[239,181,188,243,183,181,183,183,181],1.3,[239,181,191,243,183,181,183,183,181],[246,181,181,196,183,181,183,183,181],8,[246,181,185,212,183,181,183,183,181],[246,181,188,249,183,181,183,183,181],1.1,[246,181,191,241,183,181,183,183,181],[252,181,181,253,183,181,183,183,181],9,0.6,[252,181,185,220,183,181,183,183,181],[252,181,188,253,183,181,183,183,181],[252,181,191,249,183,181,183,183,181],[],[136],[],[],[262],"ICL-NUIM living-room trajectories kt0 to kt3 with synthetic noise; ATE RMSE; comparator values match those printed in ElasticFusion Table II; Ours (s) sparse-only and Ours (sd) sparse plus local dense are ablations; Redwood runs offline without colour",{"slug":264,"group":265,"sourceId":5,"sourceLabel":6,"table":266,"selfRows":137,"metrics":267,"seqs":269,"entrants":279,"cells":295,"outcomes":365,"locators":367,"hardware":368,"wordings":369,"notes":370},"bundlefusion2017-table-4","bundlefusion2017:Table 4","Table 4",[268],{"label":140,"unit":141,"statistic":142,"alignment":44},[270,273,275,277],{"dataset":113,"sequence":271,"environment":272},"fr1\u002Fdesk","small scenes with simple camera trajectories; hand-held Kinect sequences with motion-capture ground truth (Sec. 6)",{"dataset":113,"sequence":274,"environment":272},"fr2\u002Fxyz",{"dataset":113,"sequence":276,"environment":272},"fr3\u002Foffice",{"dataset":113,"sequence":278,"environment":272},"fr3\u002Fnst",[280,281,282,283,284,285,286,289,291,292,293,294],{"name":156,"methodId":100,"linkable":78,"proposed":78,"self":78},{"name":158,"methodId":100,"linkable":78,"proposed":78,"self":78},{"name":160,"methodId":100,"linkable":78,"proposed":78,"self":78},{"name":162,"methodId":163,"linkable":164,"proposed":78,"self":78},{"name":166,"methodId":167,"linkable":164,"proposed":78,"self":78},{"name":169,"methodId":170,"linkable":164,"proposed":78,"self":78},{"name":287,"methodId":288,"linkable":164,"proposed":78,"self":78},"LSD-SLAM","lsdslam2014",{"name":290,"methodId":100,"linkable":78,"proposed":78,"self":78},"Submap BA",{"name":172,"methodId":100,"linkable":78,"proposed":78,"self":78},{"name":174,"methodId":5,"linkable":164,"proposed":78,"self":164},{"name":176,"methodId":5,"linkable":164,"proposed":78,"self":164},{"name":178,"methodId":5,"linkable":164,"proposed":164,"self":164},[296,298,299,301,302,304,305,307,309,311,312,314,316,318,319,320,322,323,325,326,328,329,330,331,333,334,336,337,338,339,340,341,342,344,346,347,349,351,352,353,354,356,357,359,360,362,363,364],[181,181,181,297,183,181,183,183,181],2.1,[181,181,185,198,183,181,183,183,181],[181,181,188,300,183,181,183,183,181],3.5,[181,181,191,198,183,181,183,183,181],[185,181,181,303,183,181,183,183,181],2.3,[185,181,185,196,183,181,183,183,181],[185,181,188,306,183,181,183,183,181],3.2,[185,181,191,308,183,181,183,183,181],1.7,[188,181,181,310,183,181,183,183,181],4.3,[188,181,185,188,183,181,183,183,181],[188,181,188,313,183,181,183,183,181],4.2,[188,181,191,315,183,181,183,183,181],201.8,[191,181,181,317,183,181,183,183,181],3.7,[191,181,185,186,183,181,183,183,181],[191,181,188,191,183,181,183,183,181],[191,181,191,321,183,181,183,183,181],3.1,[217,181,181,303,183,181,183,183,181],[217,181,185,324,183,181,183,183,181],2.2,[217,181,188,303,183,181,183,183,181],[217,181,191,327,183,181,183,183,181],8.7,[224,181,181,188,183,181,183,183,181],[224,181,185,249,183,181,183,183,181],[224,181,188,308,183,181,183,183,181],[224,181,191,332,183,181,183,183,181],1.6,[231,181,181,100,181,181,183,183,181],[231,181,185,335,183,181,183,183,181],1.5,[231,181,188,100,181,181,183,183,181],[231,181,191,100,181,181,183,183,181],[239,181,181,324,183,181,183,183,181],[239,181,185,100,181,181,183,183,181],[239,181,188,300,183,181,183,183,181],[239,181,191,100,181,181,183,183,181],[246,181,181,343,183,181,183,183,181],2.7,[246,181,185,345,183,181,183,183,181],9.1,[246,181,188,191,183,181,183,183,181],[246,181,191,348,183,181,183,183,181],192.9,[252,181,181,350,183,181,183,183,181],1.9,[252,181,185,218,183,181,183,183,181],[252,181,188,186,183,181,183,183,181],[252,181,191,332,183,181,183,183,181],[355,181,181,308,183,181,183,183,181],10,[355,181,185,218,183,181,183,183,181],[355,181,188,358,183,181,183,183,181],2.8,[355,181,191,218,183,181,183,183,181],[361,181,181,332,183,181,183,183,181],11,[361,181,185,249,183,181,183,183,181],[361,181,188,324,183,181,183,183,181],[361,181,191,241,183,181,183,183,181],[366],"no value in source ('-')",[266],[],[],[371],"TUM RGB-D ATE RMSE; ground truth from a calibrated motion capture system for hand-held Kinect sequences; for Kinect data the dense reprojection threshold is 0.3 m and residuals above 0.16 m are pruned; Redwood offline and geometry-only",{"slug":373,"group":374,"sourceId":375,"sourceLabel":376,"table":136,"selfRows":246,"metrics":377,"seqs":384,"entrants":395,"cells":405,"outcomes":478,"locators":479,"hardware":480,"wordings":481,"notes":482},"badslam2019-table-3","badslam2019:Table 3","badslam2019","Schöps et al., 2019",[378,381],{"label":379,"unit":141,"statistic":380,"alignment":44},"ATE RMSE [cm], average over seven synthetic datasets","mean",{"label":382,"unit":141,"statistic":383,"alignment":44},"ATE RMSE [cm], median over seven synthetic datasets","median",[385,389,391,393],{"dataset":386,"sequence":387,"environment":388},"synthetic TUM RGB-D renders (7 datasets per category)","clean","synthetic renders of dense TUM RGB-D scene reconstructions along the original trajectories (Sec. 5)",{"dataset":386,"sequence":390,"environment":388},"async",{"dataset":386,"sequence":392,"environment":388},"rs",{"dataset":386,"sequence":394,"environment":388},"async & rs",[396,397,398,400,403],{"name":7,"methodId":5,"linkable":164,"proposed":78,"self":164},{"name":156,"methodId":100,"linkable":78,"proposed":78,"self":78},{"name":399,"methodId":170,"linkable":164,"proposed":78,"self":78},"ElasticFusion",{"name":401,"methodId":402,"linkable":164,"proposed":78,"self":78},"ORB-SLAM2","orbslam2_2017",{"name":404,"methodId":375,"linkable":164,"proposed":164,"self":78},"BAD SLAM (Ours)",[406,408,410,412,414,416,418,420,421,423,425,426,428,430,431,432,434,436,438,439,441,442,444,445,447,449,451,453,455,457,459,461,463,465,467,469,470,472,474,476],[181,181,181,407,183,181,183,183,181],0.34,[185,181,181,409,183,181,183,183,181],0.32,[188,181,181,411,183,181,183,183,181],1.11,[191,181,181,413,183,181,183,183,181],0.47,[217,181,181,415,183,181,183,183,181],0.15,[181,185,181,417,183,181,183,183,181],0.22,[185,185,181,419,183,181,183,183,181],0.23,[188,185,181,225,183,181,183,183,181],[191,185,181,422,183,181,183,183,181],0.3,[217,185,181,424,183,181,183,183,181],0.02,[181,181,185,249,183,181,183,183,181],[185,181,185,427,183,181,183,183,181],2.33,[188,181,185,429,183,181,183,183,181],1.98,[191,181,185,253,183,181,183,183,181],[217,181,185,220,183,181,183,183,181],[181,185,185,433,183,181,183,183,181],1.14,[185,185,185,435,183,181,183,183,181],0.72,[188,185,185,437,183,181,183,183,181],1.17,[191,185,185,220,183,181,183,183,181],[217,185,185,440,183,181,183,183,181],0.21,[181,181,188,249,183,181,183,183,181],[185,181,188,443,183,181,183,183,181],5.1,[188,181,188,343,183,181,183,183,181],[191,181,188,446,183,181,183,183,181],3.25,[217,181,188,448,183,181,183,183,181],0.99,[181,185,188,450,183,181,183,183,181],1.02,[185,185,188,452,183,181,183,183,181],1.37,[188,185,188,454,183,181,183,183,181],1.77,[191,185,188,456,183,181,183,183,181],1.57,[217,185,188,458,183,181,183,183,181],0.87,[181,181,191,460,183,181,183,183,181],1.48,[185,181,191,462,183,181,183,183,181],4.94,[188,181,191,464,183,181,183,183,181],3.19,[191,181,191,466,183,181,183,183,181],3.49,[217,181,191,468,183,181,183,183,181],1.01,[181,185,191,218,183,181,183,183,181],[185,185,191,471,183,181,183,183,181],1.39,[188,185,191,473,183,181,183,183,181],2.52,[191,185,191,475,183,181,183,183,181],1.55,[217,185,191,477,183,181,183,183,181],0.98,[],[136],[],[],[483],"Synthetic renders of dense TUM RGB-D reconstructions along the original trajectories; each value aggregates ATE RMSE over seven synthetic datasets per category (avg. or med.); rs uses Kinect v1 shutter times (about 30.5 ms depth, 26.1 ms colour); async renders colour midway between depth frames",{"slug":485,"group":486,"sourceId":487,"sourceLabel":488,"table":489,"selfRows":246,"metrics":490,"seqs":493,"entrants":499,"cells":509,"outcomes":531,"locators":532,"hardware":533,"wordings":534,"notes":535},"flashfusion2018-table-i","flashfusion2018:Table I","flashfusion2018","Han & Fang, 2018","Table I",[491],{"label":492,"unit":141,"statistic":142,"alignment":44},"ATE rmse (cm)",[494,496,497,498],{"dataset":113,"sequence":271,"environment":495},"real indoor office scenes, RGB-D camera (carrying mode not stated in the paper)",{"dataset":113,"sequence":274,"environment":495},{"dataset":113,"sequence":276,"environment":495},{"dataset":113,"sequence":278,"environment":495},[500,502,503,505,507],{"name":501,"methodId":100,"linkable":78,"proposed":78,"self":78},"RGBD SLAM [3] (Endres et al.)",{"name":399,"methodId":170,"linkable":164,"proposed":78,"self":78},{"name":504,"methodId":5,"linkable":164,"proposed":78,"self":164},"BundleFusion (on-line)",{"name":506,"methodId":5,"linkable":164,"proposed":78,"self":164},"BundleFusion (off-line)",{"name":508,"methodId":487,"linkable":164,"proposed":164,"self":78},"FlashFusion",[510,511,512,513,514,515,516,517,518,519,520,521,522,523,524,526,527,528,529,530],[181,181,181,303,183,181,183,183,181],[185,181,181,188,183,181,183,183,181],[188,181,181,308,183,181,183,183,181],[191,181,181,332,183,181,183,183,181],[217,181,181,350,183,181,183,183,181],[181,181,185,196,183,181,183,183,181],[185,181,185,249,183,181,183,183,181],[188,181,185,218,183,181,183,183,181],[191,181,185,249,183,181,183,183,181],[217,181,185,243,183,181,183,183,181],[181,181,188,306,183,181,183,183,181],[185,181,188,308,183,181,183,183,181],[188,181,188,186,183,181,183,183,181],[191,181,188,324,183,181,183,183,181],[217,181,188,525,183,181,183,183,181],2.5,[181,181,191,308,183,181,183,183,181],[185,181,191,332,183,181,183,183,181],[188,181,191,332,183,181,183,183,181],[191,181,191,241,183,181,183,183,181],[217,181,191,198,183,181,183,183,181],[],[489],[],[],[536],"Localization accuracy on TUM RGB-D as ATE RMSE (Sturm et al.) in cm; alignment not stated",[538,543,548,553,559,564,569,576,582,587,593],{"group":539,"slug":540,"sourceLabel":488,"table":541,"selfRows":246,"datasets":542},"flashfusion2018:Table II","flashfusion2018-table-ii","Table II",[145],{"group":544,"slug":545,"sourceLabel":6,"table":546,"selfRows":224,"datasets":547},"bundlefusion2017:Text Sec. 6 Memory","bundlefusion2017-text-sec-6-memory","Text Sec. 6 Memory",[93,123],{"group":549,"slug":550,"sourceLabel":6,"table":551,"selfRows":217,"datasets":552},"bundlefusion2017:Table 6","bundlefusion2017-table-6","Table 6",[145],{"group":554,"slug":555,"sourceLabel":6,"table":556,"selfRows":217,"datasets":557},"bundlefusion2017:Table 7","bundlefusion2017-table-7","Table 7",[558],"Augmented ICL-NUIM",{"group":560,"slug":561,"sourceLabel":562,"table":489,"selfRows":217,"datasets":563},"densesurfelmapping2019:Table I","densesurfelmapping2019-table-i","Wang et al., 2019",[145],{"group":565,"slug":566,"sourceLabel":488,"table":567,"selfRows":217,"datasets":568},"flashfusion2018:Table III","flashfusion2018-table-iii","Table III",[145],{"group":570,"slug":571,"sourceLabel":572,"table":573,"selfRows":217,"datasets":574},"photoslam2024:Table 1","photoslam2024-table-1","Huang et al., 2024c","Table 1",[575],"Replica",{"group":577,"slug":578,"sourceLabel":579,"table":580,"selfRows":217,"datasets":581},"rtgslam2024:Supp. Table 11","rtgslam2024-supp-table-11","Peng et al., 2024","Supp. Table 11",[575],{"group":583,"slug":584,"sourceLabel":376,"table":585,"selfRows":191,"datasets":586},"badslam2019:Table 2","badslam2019-table-2","Table 2",[113],{"group":588,"slug":589,"sourceLabel":590,"table":585,"selfRows":191,"datasets":591},"loopyslam2024:Table 2","loopyslam2024-table-2","Liso et al., 2024",[592],"TUM-RGBD",{"group":594,"slug":595,"sourceLabel":596,"table":597,"selfRows":188,"datasets":598},"droidslam2021:Fig. 4 table","droidslam2021-fig-4-table","Teed & Deng, 2021","Fig. 4 table",[599],"ETH3D SLAM benchmark",1790510656058]