[{"data":1,"prerenderedAt":427},["ShallowReactive",2],{"method-m2dp2016":3},{"method":4,"reference":47,"equipment":66,"figures":89,"results":90},{"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":22,"limitations":27,"sensors":32,"platform":34,"estimator":38,"association":39,"timeModel":38,"deskew":18,"loopClosure":40,"globalOptimization":41,"mapRepresentation":38,"prior":42,"outputGeometry":38,"compute":43,"codeUrl":44,"codeLicense":45,"relatedVersions":46},"m2dp2016","He et al., 2016","M2DP","M2DP: A novel 3D point cloud descriptor and its application in loop closure detection",2016,"recent","C06","place_recognition_component","M2DP 先以質心平移並用 PCA 主軸對齊點雲，再將點雲投影到 4 個方位角乘 16 個仰角共 64 個 2D 平面；每個平面以 8 個同心圓乘 16 個扇區計算點數，組成 64×128 的簽章矩陣，最後以奇異值分解取第一左、右奇異向量，得到 192 維全域描述子，用於光達迴圈偵測。作者在 KITTI00 上報告其描述子計算時間為各 3D 描述子中最短；在 KITTI、Freiburg Campus 與 Ford Campus 的多數序列中，其 100% 精確率下的召回率最高，但在 KITTI00 與 KITTI05 上 SHOT 略高，視覺描述子 GIST 更高。","M2DP builds a global descriptor from multi-plane projection density signatures compressed by SVD and applies it to loop closure detection.","full_text_reviewed","peer_reviewed_published","background","not_reported",[20,21],"public_benchmark","cross_site",[23,24,25,26],"Fastest descriptor computation among the compared 3D descriptors on KITTI00: 0.3598 s per cloud versus 0.4943 to 1.5279 s (Table II)","Highest recall at 100% precision among 3D descriptors on KITTI06 (0.899), KITTI07 (0.510), Freiburg Campus (1) and Ford Campus (0.838); on KITTI00 and KITTI05 SHOT, computed as a global descriptor about the centroid, was higher (0.898 vs 0.896; 0.816 vs 0.764) and visual GIST highest (Table I)","Robust to downsampling: recall at 100% precision 0.996, 0.996, 0.949 and 0.718 at grid sizes res x 10, 20, 50, 100 on KITTI06, while SHOT fell from 0.945 to 0.087 (Table III)","Authors report robustness to added uniform noise on KITTI07 (Fig. 5, qualitative reading only)",[28,29,30,31],"Rotation invariance relies on PCA alignment that assumes each cloud has two dominant directions (Sec. III-B)","(inference) Parameters l, t, p, q were optimised on KITTI07, which is also one of the evaluation sequences (Sec. IV-A)","Found in follow-up work: Scan Context authors report M2DP failed to detect reverse loops, e.g. KITTI 08 (scancontext2018 Sec. IV-B); Scan Context++ likewise reports failure on KITTI 08 under severe rotational variance (scancontextpp2022 Sec. VII-B1)","Found in follow-up work: STD authors report M2DP performing poorly on their Livox solid-state datasets under their setup (std2023 Sec. IV-B1)",[33],"3D LiDAR",[35,36,37],"KITTI sequences 00, 05, 06, 07 (platform not described in the paper)","Freiburg Campus dataset (77 clouds; platform not described in the paper)","Ford Campus dataset (3817 clouds; platform not described in the paper)","not_applicable","centroid-shifted, PCA-aligned cloud projected onto p x q = 4 x 16 planes (azimuth stride pi\u002Fp, elevation stride pi\u002F(2q)); each plane split into l = 8 concentric rings (radii r, 4r, ..., l^2 r) x t = 16 sectors holding point counts; 64 x 128 signature matrix reduced by SVD to the first left and right singular vectors (192-D); loop decided by thresholding the nearest-neighbour descriptor distance with +\u002F-50 frames excluded","descriptor used for loop closure detection","none (component)","none","laptop with Intel quad-core 2.50 GHz CPU and 8 GB RAM; M2DP implemented in Matlab and C; 0.3598 s per cloud to compute the descriptor and 0.0057 s for nearest-neighbour search on KITTI00 with raw clouds (Table II)",null,"not_verified",[],{"id":5,"kind":48,"shortName":7,"title":8,"authors":49,"year":9,"venue":53,"venueType":54,"publisher":55,"volumeIssuePages":56,"doi":57,"arxivId":44,"url":58,"firstPublicDate":59,"publicationStatus":16,"metadataStatus":60,"fulltextStatus":15,"era":10,"classicReason":38,"codeUrl":44,"cluster":11,"topics":61,"mdpi":62,"verification":63,"label":6,"fulltextRoute":64,"versionRead":65,"addedByCensus":62},"component",[50,51,52],"Li He","Xiaolong Wang","Hong Zhang","2016 IEEE\u002FRSJ International Conference on Intelligent Robots and Systems (IROS)","conference","IEEE","pp. 231-237","10.1109\u002Firos.2016.7759060","https:\u002F\u002Fdoi.org\u002F10.1109\u002FIROS.2016.7759060","2016-10","metadata_verified",[11],false,"corrected","NTU institutional (Chrome)","Version of record, IEEE Xplore HTML full text (IROS 2016, document 7759060); Tables I to III read from the publisher table images",[67,73,79,82,85],{"category":68,"model":69,"canonical":69,"role":70,"dataset":44,"specs":71,"locator":72},"compute","laptop, Intel quad-core 2.50 GHz CPU, 8 GB RAM","compute for runtime","quad-core 2.50 GHz, 8 GB RAM","Sec. IV",{"category":74,"model":75,"canonical":75,"role":76,"dataset":77,"specs":18,"locator":78},"lidar","not_reported (KITTI lidar sequences 00, 05, 06, 07)","dataset sensor","KITTI","Sec. IV-A",{"category":74,"model":80,"canonical":80,"role":76,"dataset":81,"specs":18,"locator":78},"not_reported (Freiburg Campus lidar data, 77 point clouds)","Freiburg Campus",{"category":74,"model":83,"canonical":83,"role":76,"dataset":84,"specs":18,"locator":78},"not_reported (Ford Campus lidar data, 3817 point clouds)","Ford Campus",{"category":86,"model":87,"canonical":87,"role":76,"dataset":88,"specs":18,"locator":78},"camera","not_reported (dataset images used for the GIST visual baseline)","KITTI; Ford Campus",[],{"totalRows":91,"groupCount":92,"groups":93,"others":417},21,6,[94,222,294,351],{"slug":95,"group":96,"sourceId":5,"sourceLabel":6,"table":97,"selfRows":92,"metrics":98,"seqs":102,"entrants":117,"cells":132,"outcomes":215,"locators":217,"hardware":218,"wordings":219,"notes":220},"m2dp2016-table-i","m2dp2016:Table I","Table I",[99],{"label":100,"unit":101,"statistic":18,"alignment":38},"recall at 100% precision","ratio",[103,106,108,110,112,115],{"dataset":77,"sequence":104,"environment":105},"00","urban driving",{"dataset":77,"sequence":107,"environment":105},"05",{"dataset":77,"sequence":109,"environment":105},"06",{"dataset":77,"sequence":111,"environment":105},"07",{"dataset":81,"sequence":113,"environment":114},"all (77 clouds)","campus",{"dataset":84,"sequence":116,"environment":114},"all (3817 clouds)",[118,120,122,124,126,128,130],{"name":119,"methodId":44,"linkable":62,"proposed":62,"self":62},"Spin Image",{"name":121,"methodId":44,"linkable":62,"proposed":62,"self":62},"SHOT",{"name":123,"methodId":44,"linkable":62,"proposed":62,"self":62},"VFH",{"name":125,"methodId":44,"linkable":62,"proposed":62,"self":62},"ESF",{"name":127,"methodId":44,"linkable":62,"proposed":62,"self":62},"Z-projection",{"name":7,"methodId":5,"linkable":129,"proposed":129,"self":129},true,{"name":131,"methodId":44,"linkable":62,"proposed":62,"self":62},"GIST",[133,137,140,142,145,148,151,153,155,157,158,160,162,164,166,168,170,172,174,176,178,180,181,183,184,186,188,189,191,193,195,196,198,200,201,202,204,206,207,209,211,213],[134,134,134,135,136,134,136,136,134],0,0.271,-1,[138,134,134,139,136,134,136,136,134],1,0.898,[141,134,134,134,136,134,136,136,134],2,[143,134,134,144,136,134,136,136,134],3,0.174,[146,134,134,147,136,134,136,136,134],4,0.456,[149,134,134,150,136,134,136,136,134],5,0.896,[92,134,134,152,136,134,136,136,134],0.957,[134,134,138,154,136,134,136,136,134],0.51,[138,134,138,156,136,134,136,136,134],0.816,[141,134,138,134,136,134,136,136,134],[143,134,138,159,136,134,136,136,134],0.557,[146,134,138,161,136,134,136,136,134],0.513,[149,134,138,163,136,134,136,136,134],0.764,[92,134,138,165,136,134,136,136,134],0.855,[134,134,141,167,136,134,136,136,134],0.606,[138,134,141,169,136,134,136,136,134],0.666,[141,134,141,171,136,134,136,136,134],0.029,[143,134,141,173,136,134,136,136,134],0.233,[146,134,141,175,136,134,136,136,134],0.737,[149,134,141,177,136,134,136,136,134],0.899,[92,134,141,179,136,134,136,136,134],0.885,[134,134,143,134,136,134,136,136,134],[138,134,143,182,136,134,136,136,134],0.466,[141,134,143,134,136,134,136,136,134],[143,134,143,185,136,134,136,136,134],0.098,[146,134,143,187,136,134,136,136,134],0.383,[149,134,143,154,136,134,136,136,134],[92,134,143,190,136,134,136,136,134],0.451,[134,134,146,192,136,134,136,136,134],0.6,[138,134,146,194,136,134,136,136,134],0.769,[141,134,146,134,136,134,136,136,134],[143,134,146,197,136,134,136,136,134],0.363,[146,134,146,199,136,134,136,136,134],0.714,[149,134,146,138,136,134,136,136,134],[92,134,146,44,134,134,136,136,134],[134,134,149,203,136,134,136,136,134],0.177,[138,134,149,205,136,134,136,136,134],0.798,[141,134,149,134,136,134,136,136,134],[143,134,149,208,136,134,136,136,134],0.774,[146,134,149,210,136,134,136,136,134],0.702,[149,134,149,212,136,134,136,136,134],0.838,[92,134,149,214,136,134,136,136,134],0.789,[216],"not_run (no images in Freiburg Campus)",[97],[],[],[221],"Recall at 100% precision, raw clouds, nearest-neighbour matching, loop if GT distance \u003C 10 m, +\u002F-50 frames excluded (+\u002F-5 Freiburg)",{"slug":223,"group":224,"sourceId":5,"sourceLabel":6,"table":225,"selfRows":146,"metrics":226,"seqs":228,"entrants":237,"cells":244,"outcomes":288,"locators":289,"hardware":290,"wordings":291,"notes":292},"m2dp2016-table-iii","m2dp2016:Table III","Table III",[227],{"label":100,"unit":101,"statistic":18,"alignment":38},[229,231,233,235],{"dataset":77,"sequence":230,"environment":105},"06 (grid size res x10)",{"dataset":77,"sequence":232,"environment":105},"06 (grid size res x20)",{"dataset":77,"sequence":234,"environment":105},"06 (grid size res x50)",{"dataset":77,"sequence":236,"environment":105},"06 (grid size res x100)",[238,239,240,241,242,243],{"name":119,"methodId":44,"linkable":62,"proposed":62,"self":62},{"name":121,"methodId":44,"linkable":62,"proposed":62,"self":62},{"name":123,"methodId":44,"linkable":62,"proposed":62,"self":62},{"name":125,"methodId":44,"linkable":62,"proposed":62,"self":62},{"name":127,"methodId":44,"linkable":62,"proposed":62,"self":62},{"name":7,"methodId":5,"linkable":129,"proposed":129,"self":129},[245,247,249,250,252,254,256,258,260,261,263,265,266,268,270,271,273,275,277,279,281,282,284,286],[134,134,134,246,136,134,136,136,134],0.154,[138,134,134,248,136,134,136,136,134],0.945,[141,134,134,134,136,134,136,136,134],[143,134,134,251,136,134,136,136,134],0.071,[146,134,134,253,136,134,136,136,134],0.007,[149,134,134,255,136,134,136,136,134],0.996,[134,134,138,257,136,134,136,136,134],0.109,[138,134,138,259,136,134,136,136,134],0.834,[141,134,138,134,136,134,136,136,134],[143,134,138,262,136,134,136,136,134],0.037,[146,134,138,264,136,134,136,136,134],0.008,[149,134,138,255,136,134,136,136,134],[134,134,141,267,136,134,136,136,134],0.047,[138,134,141,269,136,134,136,136,134],0.415,[141,134,141,134,136,134,136,136,134],[143,134,141,272,136,134,136,136,134],0.056,[146,134,141,274,136,134,136,136,134],0.023,[149,134,141,276,136,134,136,136,134],0.949,[134,134,143,278,136,134,136,136,134],0.021,[138,134,143,280,136,134,136,136,134],0.087,[141,134,143,134,136,134,136,136,134],[143,134,143,283,136,134,136,136,134],0.005,[146,134,143,285,136,134,136,136,134],0.027,[149,134,143,287,136,134,136,136,134],0.718,[],[225],[],[],[293],"Recall at 100% precision on KITTI06 after grid downsampling with grid size res x k (res = mean nearest-neighbour distance in first frame)",{"slug":295,"group":296,"sourceId":297,"sourceLabel":298,"table":299,"selfRows":146,"metrics":300,"seqs":305,"entrants":310,"cells":321,"outcomes":344,"locators":346,"hardware":347,"wordings":348,"notes":349},"overlapnet2020-table-ii","overlapnet2020:Table II","overlapnet2020","Chen et al., 2020","Table II",[301,303],{"label":302,"unit":101,"statistic":18,"alignment":38},"AUC (precision-recall)",{"label":304,"unit":101,"statistic":18,"alignment":38},"F1 score",[306,308],{"dataset":307,"sequence":104,"environment":105},"KITTI odometry",{"dataset":84,"sequence":104,"environment":309},"campus and downtown driving",[311,313,314,317,319],{"name":312,"methodId":44,"linkable":62,"proposed":62,"self":62},"Histogram",{"name":7,"methodId":5,"linkable":129,"proposed":62,"self":129},{"name":315,"methodId":316,"linkable":129,"proposed":62,"self":62},"SuMa","suma2018",{"name":318,"methodId":297,"linkable":129,"proposed":129,"self":62},"Ours (AllChannel, TwoHeads)",{"name":320,"methodId":297,"linkable":129,"proposed":129,"self":62},"Ours (GeoOnly)",[322,324,325,326,328,329,331,332,334,336,337,338,339,340,342,343],[134,134,134,323,136,134,136,136,134],0.83,[134,138,134,323,136,134,136,136,134],[138,134,134,323,136,134,136,136,134],[138,138,134,327,136,134,136,136,134],0.87,[141,134,134,44,134,134,136,136,134],[141,138,134,330,136,134,136,136,134],0.85,[143,134,134,327,136,134,136,136,134],[143,138,134,333,136,134,136,136,134],0.88,[134,134,138,335,136,134,136,136,134],0.84,[134,138,138,323,136,134,136,136,134],[138,134,138,335,136,134,136,136,134],[138,138,138,330,136,134,136,136,134],[141,134,138,44,134,134,136,136,134],[141,138,138,341,136,134,136,136,134],0.33,[146,134,138,330,136,134,136,136,134],[146,138,138,335,136,134,136,136,134],[345],"not_applicable (SuMa yields a single precision-recall point)",[299],[],[],[350],"Loop closure detection vs state of the art; best candidate per query, 100 latest scans excluded, overlap threshold 30%; KITTI uses all cues, Ford uses geometry only",{"slug":352,"group":353,"sourceId":354,"sourceLabel":355,"table":356,"selfRows":143,"metrics":357,"seqs":365,"entrants":367,"cells":378,"outcomes":408,"locators":409,"hardware":410,"wordings":414,"notes":415},"scancontextpp2022-table-v","scancontextpp2022:Table V","scancontextpp2022","Kim et al., 2022b","Table V",[358,361,363],{"label":359,"unit":360,"statistic":18,"alignment":38},"Description time","ms",{"label":362,"unit":360,"statistic":18,"alignment":38},"Retrieval time",{"label":364,"unit":360,"statistic":18,"alignment":38},"Total time",[366],{"dataset":18,"sequence":18,"environment":18},[368,370,372,373,375],{"name":369,"methodId":354,"linkable":129,"proposed":129,"self":62},"Ours (PC)",{"name":371,"methodId":354,"linkable":129,"proposed":129,"self":62},"Ours (A-PC)",{"name":7,"methodId":5,"linkable":129,"proposed":62,"self":129},{"name":374,"methodId":44,"linkable":62,"proposed":62,"self":62},"SegMatch",{"name":376,"methodId":377,"linkable":129,"proposed":62,"self":62},"PointNetVLAD","pointnetvlad2018",[379,381,383,385,387,388,390,392,394,396,398,400,402,404,406],[134,134,134,380,136,134,134,136,134],1.6,[134,138,134,382,136,134,134,136,134],6.7,[134,141,134,384,136,134,134,136,134],8.3,[138,134,134,386,136,134,134,136,134],4.8,[138,138,134,382,136,134,134,136,134],[138,141,134,389,136,134,134,136,134],11.5,[141,134,134,391,136,134,134,136,134],4.3,[141,138,134,393,136,134,134,136,134],1.5,[141,141,134,395,136,134,134,136,134],5.8,[143,134,134,397,136,134,138,136,134],430.2,[143,138,134,399,136,134,138,136,134],365.8,[143,141,134,401,136,134,138,136,134],796,[146,134,134,403,136,134,141,136,134],33.3,[146,138,134,405,136,134,141,136,134],0.7,[146,141,134,407,136,134,141,136,134],34,[],[356],[411,412,413],"Matlab, CPU (machine not stated for Table V)","values copied from the SegMatch paper","GPU (GTX 1080 Ti)",[],[416],"Time cost per query in ms; ours and M2DP measured in Matlab, SegMatch copied from its paper, PointNetVLAD on GPU",[418,422],{"group":419,"slug":420,"sourceLabel":6,"table":299,"selfRows":141,"datasets":421},"m2dp2016:Table II","m2dp2016-table-ii",[77],{"group":423,"slug":424,"sourceLabel":425,"table":299,"selfRows":141,"datasets":426},"scancontext2018:Table II","scancontext2018-table-ii","Kim & Kim, 2018",[77],1790510663130]