| { |
| "arctic": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "光学动捕与多相机采集系统", |
| "caption": "光学动捕与多相机采集系统", |
| "source_url": "https://arxiv.org/html/2204.13662#S2.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Figure 4 : Our capture system with 54 54 high-resolution MoCap cameras .\n(a) MoCap system in a side view and a bird-eye view, illustrating the 54 54 MoCap cameras used to eliminate occlusion during the capture.\n(b) Observed markers for a captured frame, showing markers tracking the full human body with hands, the object (notebook in this case), the egocentric camera, and props such as the table.\nBest viewed in color and zoomed in.", |
| "attribution": "Zicong Fan Affiliation: ETH Zürich, Switzerland University of Amsterdam Max Planck Institute for Intelligent Systems, Tübingen, Germany Omid Taheri Affiliation: ETH Zürich, Switzerland University of Amsterdam Max Planck Institute for Intelligent Systems, Tübingen, Germany Dimitrios Tzionas Muhammed Kocabas Affiliation: ETH Zürich, Switzerland University of Amsterdam Max Planck Institute for Intelligent Systems, Tübingen, Germany Manuel Kaufmann Michael J. Black Affiliation: ETH Zürich, Switzerland University of Amsterdam Max Planck Institute for Intelligent Systems, Tübingen, Germany Otmar Hilliges", |
| "images": [ |
| { |
| "src": "media/figures/arctic_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2204.13662v3/1_figures/vicon_system.png", |
| "width": 2168, |
| "height": 591, |
| "sha256": "687c5b27bf77928208899d457e20f19e3f122ae116d7e3f3b35c1f24606809af", |
| "panel": 0 |
| } |
| ] |
| }, |
| { |
| "title": "手、物体与相机的反光标记", |
| "caption": "手、物体与相机的反光标记", |
| "source_url": "https://arxiv.org/html/2204.13662#S1.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Markers for motion capture ( MoCap ) . We put 1.5 1.5 mm radius markers on objects, hands and the egocentric camera. For the body, we use 4.5 4.5 mm radius markers. The markers are shown here to scale. Best viewed in color and zoomed-in.", |
| "attribution": "Zicong Fan Affiliation: ETH Zürich, Switzerland University of Amsterdam Max Planck Institute for Intelligent Systems, Tübingen, Germany Omid Taheri Affiliation: ETH Zürich, Switzerland University of Amsterdam Max Planck Institute for Intelligent Systems, Tübingen, Germany Dimitrios Tzionas Muhammed Kocabas Affiliation: ETH Zürich, Switzerland University of Amsterdam Max Planck Institute for Intelligent Systems, Tübingen, Germany Manuel Kaufmann Michael J. Black Affiliation: ETH Zürich, Switzerland University of Amsterdam Max Planck Institute for Intelligent Systems, Tübingen, Germany Otmar Hilliges", |
| "images": [ |
| { |
| "src": "media/figures/arctic_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2204.13662v3/1_figures/capture_setup.jpg", |
| "width": 6973, |
| "height": 1888, |
| "sha256": "e8b0c21d73c3c4c6efbbc53fe463b4b755e2617445a2954d5e2706be821edf5e", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "hot3d": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "OptiTrack 动捕实验室", |
| "caption": "OptiTrack 动捕实验室", |
| "source_url": "https://arxiv.org/html/2411.19167#S3.F5", |
| "source_label": "论文图 5", |
| "original_caption": "Figure 5 : Motion-capture lab. The HOT3D dataset was collected using a motion-capture rig equipped with a few dozens of infrared exocentric OptiTrack cameras and light diffuser panels for illumination variability.", |
| "attribution": "Prithviraj Banerjee, Sindi Shkodrani, Pierre Moulon, Shreyas Hampali, Shangchen Han Affiliation: Fan Zhang, Linguang Zhang, Jade Fountain, Edward Miller, Selen Basol, Affiliation: Richard Newcombe, Robert Wang, Jakob Julian Engel, Tomas Hodan Affiliation: Meta Reality Labs facebookresearch.github.io/hot3d", |
| "images": [ |
| { |
| "src": "media/figures/hot3d_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2411.19167v2/figures/hot3d_zurich_lab_v5_small.jpg", |
| "width": 1280, |
| "height": 1280, |
| "sha256": "aafeaa6c59641a901a0f1e1225f52d2ec5103b593b121ccd61e92a304c67ca6e", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "h2o": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "手与物体标注流程", |
| "caption": "手与物体标注流程", |
| "source_url": "https://arxiv.org/html/2104.11181#S1.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2: (a) We calibrate cameras using IR sphere markers and PnP [ 45 ] , (b) create object meshes using BADSLAM [ 68 ] on RGB-D captures, and (c) estimate object poses using DenseFusion [ 86 ] on RGB-D images and mask images from Mask R-CNN [ 35 ] . We then select the pose with the highest confidence among five cameras. (d) Consequently, we detect hand joints with OpenPose [ 8 ] and optimize hand shape using Eq. 1 . (e) We finally detect and smooth temporally inaccurate poses.", |
| "attribution": "Taein Kwon Affiliation: ETH Zurich Bugra Tekin Affiliation: Microsoft Jan Stühmer † † thanks: Work performed while\nat Microsoft. Affiliation: Samsung AI Center, Cambridge Federica Bogo Affiliation: ETH Zurich Marc Pollefeys Affiliation: ETH Zurich Affiliation: Microsoft", |
| "images": [ |
| { |
| "src": "media/figures/h2o_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2104.11181v2/figures/pipeline3.png", |
| "width": 1205, |
| "height": 250, |
| "sha256": "ff289b2a257d9174e3510f8514c2773d31aa286c61fb28b7904b97599f7b16c4", |
| "panel": 0 |
| } |
| ] |
| }, |
| { |
| "title": "多相机布置与场景", |
| "caption": "多相机布置与场景", |
| "source_url": "https://arxiv.org/html/2104.11181#S4.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Figure 4: (a) Number of instances per action in the H2O dataset. (b) Average number of frames per each action class. (c) Schematic camera capture setup. Four static cameras can capture parts that can not be observed by the egocentric view. (d) Scene point cloud computed from multi-view data.", |
| "attribution": "Taein Kwon Affiliation: ETH Zurich Bugra Tekin Affiliation: Microsoft Jan Stühmer † † thanks: Work performed while\nat Microsoft. Affiliation: Samsung AI Center, Cambridge Federica Bogo Affiliation: ETH Zurich Marc Pollefeys Affiliation: ETH Zurich Affiliation: Microsoft", |
| "images": [ |
| { |
| "src": "media/figures/h2o_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2104.11181v2/h2odataset.png", |
| "width": 1600, |
| "height": 900, |
| "sha256": "b1d4cefd650c7901c3ce8e7174351b857f0ee783e9ce86e10798ddab391f76b6", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "hoi4d": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "头戴式 RGB-D 真实采集设备", |
| "caption": "头戴式 RGB-D 真实采集设备", |
| "source_url": "https://arxiv.org/html/2203.01577#S3.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Data capturing system. we build up a simple head-mounted data capturing suite consists of a bicycle helmet, a Kinect v2 RGB-D sensor, and an Intel RealSense D455 RGB-D sensor.", |
| "attribution": "Yunze Liu Affiliation: Tsinghua University Affiliation: Shanghai Qi Zhi Institute http://hoi4d.top/ Yun Liu Affiliation: Tsinghua University Che Jiang Affiliation: Tsinghua University Kangbo Lyu Affiliation: Tsinghua University Weikang Wan Affiliation: Peking University Hao Shen Affiliation: Peking University Boqiang Liang Affiliation: Peking University Zhoujie Fu Affiliation: Tsinghua University He Wang Affiliation: Peking University Li Yi Affiliation: Tsinghua University Affiliation: Shanghai Qi Zhi Institute http://hoi4d.top/", |
| "images": [ |
| { |
| "src": "media/figures/hoi4d_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2203.01577v4/capture.png", |
| "width": 1749, |
| "height": 865, |
| "sha256": "82a1dedd90342bcf20e815c6801e5bb71be95d2265c532e1245c3677a4b0b8ad", |
| "panel": 0 |
| } |
| ] |
| }, |
| { |
| "title": "HOI4D 标注流程", |
| "caption": "HOI4D 标注流程", |
| "source_url": "https://arxiv.org/html/2203.01577#S3.F3", |
| "source_label": "论文图 3", |
| "original_caption": "Figure 3 : Overview of annotation pipeline. Red branch : Given a dynamic RGB-D sequence, we first annotate frame-wise 2D motion segmentation. Then we mask out the moving content and reconstruct a 3D static scene. We manually annotate the reconstructed scene to obtain 3D static scene panoptic segmentation. Finally, the 2D motion segmentation and the 3D static scene panoptic segmentation are merged, resulting in the 4D dynamic scene panoptic segmentation. Blue branch : To obtain 3D hand pose labels, we first annotate a set of hand keypoints on RGB-D frames and then leverage an optimization module to recover the underlying 3D hand. For category-level object poses, we manually fit amodal oriented bounding boxes to objects or object parts in RGB-D frames and further optimize it by leveraging the object mesh. Green branch : We directly annotated fine-grained action labels on the original video.", |
| "attribution": "Yunze Liu Affiliation: Tsinghua University Affiliation: Shanghai Qi Zhi Institute http://hoi4d.top/ Yun Liu Affiliation: Tsinghua University Che Jiang Affiliation: Tsinghua University Kangbo Lyu Affiliation: Tsinghua University Weikang Wan Affiliation: Peking University Hao Shen Affiliation: Peking University Boqiang Liang Affiliation: Peking University Zhoujie Fu Affiliation: Tsinghua University He Wang Affiliation: Peking University Li Yi Affiliation: Tsinghua University Affiliation: Shanghai Qi Zhi Institute http://hoi4d.top/", |
| "images": [ |
| { |
| "src": "media/figures/hoi4d_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2203.01577v4/annotation.png", |
| "width": 2041, |
| "height": 967, |
| "sha256": "534938fb67219336460e120bf3d77dad981a31acd42b30b67afebac1249aa128", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "dexycb": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "8 台 RGB-D 相机的真实布置", |
| "caption": "8 台 RGB-D 相机的真实布置", |
| "source_url": "https://arxiv.org/html/2104.04631#S2.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Our setup with 8 RGB-D cameras (red circle).", |
| "attribution": "Yu-Wei Chao 1 Wei Yang 1 Yu Xiang 1 Pavlo Molchanov 1 Ankur Handa 1 Jonathan Tremblay 1 Yashraj S. Narang 1 Karl Van Wyk 1 Umar Iqbal 1 Stan Birchfield 1 Jan Kautz 1 Dieter Fox 1,2 1 NVIDIA 2 University of Washington {ychao,weiy,yux,pmolchanov,ahanda,jtremblay,ynarang,kvanwyk,uiqbal,sbirchfield,jkautz,dieterf}@nvidia.com", |
| "images": [ |
| { |
| "src": "media/figures/dexycb_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2104.04631v1/setup.png", |
| "width": 2596, |
| "height": 1860, |
| "sha256": "fa7381982440740eecde345956427f7a7b0f9eeb7274b75f7a0f25eae084ad66", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "ho3d": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "多相机与单相机标注流程", |
| "caption": "多相机与单相机标注流程", |
| "source_url": "https://arxiv.org/html/1907.01481#S3.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : The different stages of the multi-camera and single camera setups. See Section 4 for more details.", |
| "attribution": "Shreyas Hampali Affiliation: Institute for Computer Graphics and Vision, Graz University of Technology, Austria Mahdi Rad Affiliation: Institute for Computer Graphics and Vision, Graz University of Technology, Austria Markus Oberweger Affiliation: Institute for Computer Graphics and Vision, Graz University of Technology, Austria Vincent Lepetit Affiliation: { hampali, rad, oberweger, lepetit } @icg.tugraz.at Project Page: https://www.tugraz.at/index.php?id=40231 Affiliation: Institute for Computer Graphics and Vision, Graz University of Technology, Austria Affiliation: LIGM, Ecole des Ponts, Univ Gustave Eiffel, CNRS, Marne-la-Vallée, France", |
| "images": [ |
| { |
| "src": "media/figures/ho3d_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/1907.01481v6/singleview_pipeline.png", |
| "width": 1323, |
| "height": 745, |
| "sha256": "d98b9c9cac5fdc67ee9c3e38db08ce06271e0a42fc099fc4f8a6a3e385e52c9d", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "h2o3d": { |
| "note": "作者发布了数据样例，论文说明使用 5 台 RGB-D 相机采集；此处将数据示例与采集装置图分别标明。", |
| "figures": [ |
| { |
| "title": "作者发布的 H₂O-3D 数据样例", |
| "caption": "作者发布的 H₂O-3D 数据样例", |
| "source_url": "https://arxiv.org/html/2104.14639#S10.F15", |
| "source_label": "论文图 15", |
| "original_caption": "Figure 15 : Samples from H 2 O-3D dataset. Our dataset contains sequences with complex actions performed by both hands on YCB [ 64 ] objects.", |
| "attribution": "Shreyas Hampali Affiliation: Institute for Computer Graphics and Vision, Graz University of Technology, Graz, Austria Sayan Deb Sarkar Affiliation: Institute for Computer Graphics and Vision, Graz University of Technology, Graz, Austria Mahdi Rad Affiliation: Institute for Computer Graphics and Vision, Graz University of Technology, Graz, Austria Vincent Lepetit Email: {<firstname>.<lastname>}@icg.tugraz.at, vincent.lepetit@enpc.frProject Website: https://www.tugraz.at/index.php?id=57823 Affiliation: Institute for Computer Graphics and Vision, Graz University of Technology, Graz, Austria Affiliation: LIGM, Ecole des Ponts, Univ Gustave Eiffel, CNRS, Marne-la-Vallée, France", |
| "images": [ |
| { |
| "src": "media/figures/h2o3d_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2104.14639v2/h2o3d_vis_hor.png", |
| "width": 1322, |
| "height": 744, |
| "sha256": "3cbe0c1eb9242ffb627b32eaa8caa40b6aa9b4ae35e9d7a783cdc4c42376f57f", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "interhand": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "作者的多视角采集棚", |
| "caption": "作者的多视角采集棚", |
| "source_url": "https://arxiv.org/html/2008.09309#S12.F14", |
| "source_label": "论文图 14", |
| "original_caption": "Figure 14 : Rendering of our constructed multi-view studio.", |
| "attribution": "Gyeongsik Moon Affiliation: ECE & ASRI, Seoul National University, Korea Shoou-I Yu Affiliation: Facebook Reality Labs , E-mail {mks0601,kyoungmu}@snu.ac.kr He Wen Affiliation: Facebook Reality Labs , E-mail {mks0601,kyoungmu}@snu.ac.kr Takaaki Shiratori Affiliation: Facebook Reality Labs , E-mail {mks0601,kyoungmu}@snu.ac.kr Kyoung Mu Lee E-mail {shoou-i.yu,hewen,tshiratori}@fb.com Affiliation: ECE & ASRI, Seoul National University, Korea", |
| "images": [ |
| { |
| "src": "media/figures/interhand_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2008.09309v1/studio.png", |
| "width": 807, |
| "height": 602, |
| "sha256": "62875fc5105ace25dd80f1a7edbad22217c19c144718cbf2243e20ae99697a1e", |
| "panel": 0 |
| } |
| ] |
| }, |
| { |
| "title": "人工关键点标注流程", |
| "caption": "人工关键点标注流程", |
| "source_url": "https://arxiv.org/html/2008.09309#S11.F13", |
| "source_label": "论文图 13", |
| "original_caption": "Figure 13 : The human annotation procedure of InterHand2.6M.", |
| "attribution": "Gyeongsik Moon Affiliation: ECE & ASRI, Seoul National University, Korea Shoou-I Yu Affiliation: Facebook Reality Labs , E-mail {mks0601,kyoungmu}@snu.ac.kr He Wen Affiliation: Facebook Reality Labs , E-mail {mks0601,kyoungmu}@snu.ac.kr Takaaki Shiratori Affiliation: Facebook Reality Labs , E-mail {mks0601,kyoungmu}@snu.ac.kr Kyoung Mu Lee E-mail {shoou-i.yu,hewen,tshiratori}@fb.com Affiliation: ECE & ASRI, Seoul National University, Korea", |
| "images": [ |
| { |
| "src": "media/figures/interhand_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2008.09309v1/human_annotation.png", |
| "width": 1170, |
| "height": 363, |
| "sha256": "ab665f3accfc9b457a876468f851bc7618418fee81b92f779d4a0577cf402c68", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "reinterhand": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "采集、几何跟踪与重光照生成流程", |
| "caption": "采集、几何跟踪与重光照生成流程", |
| "source_url": "https://arxiv.org/html/2310.17768#S2.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Figure 4: The overall pipeline of constructing our Re:InterHand dataset.", |
| "attribution": "Gyeongsik Moon Email: mks0601@meta.com Shunsuke Saito Email: shunsukesaito@meta.com Weipeng Xu Email: xuweipeng@meta.com Rohan Joshi Email: rohanjoshi@meta.com Julia Buffalini Email: jbuffalini@meta.com Harley Bellan Email: harleybellan@meta.com Nicholas Rosen Email: nicholasrosen@meta.com Jesse Richardson Email: jesserichardson@meta.com Mallorie Mize Email: malloriemize@meta.com Philippe de Bree phillippedebree@meta.com Tomas Simon tsimon@meta.com Bo Peng bopeng@meta.com Shubham Garg ssgarg@meta.com Kevyn McPhail kmcphail@meta.com Takaaki Shiratori tshiratori@meta.com Meta Reality Labs Research", |
| "images": [ |
| { |
| "src": "media/figures/reinterhand_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2310.17768v1/dataset_construction_pipeline.png", |
| "width": 825, |
| "height": 540, |
| "sha256": "721d523fbd2016f5cf0bbfe2bddbd4526a90a515b06710fab067e5133ac7facc", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "freihand": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "8 相机绿幕采集装置", |
| "caption": "8 相机绿幕采集装置", |
| "source_url": "https://arxiv.org/html/1909.04349#S3.F3", |
| "source_label": "论文图 3", |
| "original_caption": "Figure 3 : Recording setup with 8 8 calibrated and temporally synchronized RGB cameras located at the corners of a cube. A green screen background can be mounted into the the setup, enabling easier background subtraction.", |
| "attribution": "Christian Zimmermann Affiliation: University of Freiburg Duygu Ceylan Affiliation: Adobe Research Jimei Yang Affiliation: Adobe Research Bryan Russell Affiliation: Adobe Research Max Argus Affiliation: University of Freiburg Thomas Brox Affiliation: Project page: https://lmb.informatik.uni-freiburg.de/projects/freihand/ Affiliation: University of Freiburg", |
| "images": [ |
| { |
| "src": "media/figures/freihand_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/1909.04349v3/figures/setup.jpg", |
| "width": 1999, |
| "height": 1500, |
| "sha256": "122335b4eaf2e300abb89d212b6d33698d4eb8e81749befaff3eba94d8390f6b", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "mtc": { |
| "note": "此处展示原论文的数据图像与三维标注。", |
| "figures": [ |
| { |
| "title": "原始多视角图像与三维标注", |
| "caption": "原始多视角图像与三维标注", |
| "source_url": "https://arxiv.org/html/1812.01598#A1.F10", |
| "source_label": "论文图 10", |
| "original_caption": "Figure 10 : Example images and 3D annotations from our new 3D human pose dataset.", |
| "attribution": "Donglai Xiang Hanbyul Joo Yaser Sheikh Affiliation: Carnegie Mellon University Email: {donglaix,hanbyulj,yaser}@cs.cmu.edu", |
| "images": [ |
| { |
| "src": "media/figures/mtc_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/1812.01598v1/image_171026_pose1_00_00_00018000.jpg", |
| "width": 1920, |
| "height": 1080, |
| "sha256": "8b38d2e9767f4dd24447bc19be4542ec5d6d435103013d668e4491839a61ff2a", |
| "panel": 0 |
| }, |
| { |
| "src": "media/figures/mtc_0_1.webp", |
| "source_image_url": "https://arxiv.org/html/1812.01598v1/body_171026_pose1_00_00_00018000.jpg", |
| "width": 1920, |
| "height": 1080, |
| "sha256": "7dbef04207c6166f81f787836c8698a11c01b1e072bfe6d1fc44aa5328616c15", |
| "panel": 1 |
| }, |
| { |
| "src": "media/figures/mtc_0_2.webp", |
| "source_image_url": "https://arxiv.org/html/1812.01598v1/image_171204_pose2_00_01_00026331.jpg", |
| "width": 1920, |
| "height": 1080, |
| "sha256": "a594e587a9672893e2ca2b0f5added2d3c5cf166364e0a3cfc380059f5c0bc60", |
| "panel": 2 |
| }, |
| { |
| "src": "media/figures/mtc_0_3.webp", |
| "source_image_url": "https://arxiv.org/html/1812.01598v1/body_171204_pose2_00_01_00026331.jpg", |
| "width": 1920, |
| "height": 1080, |
| "sha256": "46973dcca2fb029d369e0188fb613ce8dd1195e4a7a592f7aade8797e018e3d7", |
| "panel": 3 |
| } |
| ] |
| } |
| ] |
| }, |
| "bedlam": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "人体、服装、动作与场景的合成过程", |
| "caption": "人体、服装、动作与场景的合成过程", |
| "source_url": "https://arxiv.org/html/2306.16940#S2.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Dataset construction. Illustration of each step in the process, shown for a single character.\nLeft to right:\n(a) sampled body shape.\n(b) skin texture.\n(c) clothing simulation.\n(d) cloth texture.\n(e) hair.\n(f) pose.\n(g) scene and illumination.\n(h) motion blur.", |
| "attribution": "Michael J. Black 1 1 1 The authors contributed equally and are listed alphabetically. Affiliation: Max Planck Institute for Intelligent Systems, Tübingen, Germany Priyanka Patel 1 1 1 The authors contributed equally and are listed alphabetically. Affiliation: Max Planck Institute for Intelligent Systems, Tübingen, Germany Joachim Tesch 1 1 1 The authors contributed equally and are listed alphabetically. Affiliation: Max Planck Institute for Intelligent Systems, Tübingen, Germany Jinlong Yang 1 1 1 The authors contributed equally and are listed alphabetically. 2 2 2 This work was performed when JY was at MPI-IS. Affiliation: Google", |
| "images": [ |
| { |
| "src": "media/figures/bedlam_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2306.16940v1/Figures/Images/augmentation_pipeline_2000.png", |
| "width": 2000, |
| "height": 269, |
| "sha256": "fabf7482964cd3fd53233e9ff1f2b42c385c57c4111dc356db723c99204541db", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "rhd": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "合成 RGB、分割与关键点标注", |
| "caption": "合成 RGB、分割与关键点标注", |
| "source_url": "https://arxiv.org/html/1705.01389#S5.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Figure 4 : Our new dataset provides segmentation maps with 33 33 classes: three for each finger, palm, person, and background. The 3D kinematic model of the hand provides 21 21 keypoints per hand: 4 keypoints per finger and one keypoint close to the wrist.", |
| "attribution": "Christian Zimmermann Affiliation: University of Freiburg Thomas Brox Affiliation: {zimmermann, brox}@cs.uni-freiburg.de", |
| "images": [ |
| { |
| "src": "media/figures/rhd_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/1705.01389v3/figures/210image_rgb_w_kp.png", |
| "width": 320, |
| "height": 320, |
| "sha256": "4ce9b60b0245d34ff6589ac4826dc2663afe789f102d55d27fdb473cab1e7e1a", |
| "panel": 0 |
| }, |
| { |
| "src": "media/figures/rhd_0_1.webp", |
| "source_image_url": "https://arxiv.org/html/1705.01389v3/figures/210seg.png", |
| "width": 320, |
| "height": 320, |
| "sha256": "2fa108e87eccf75036960f0086432b61bebc6805236762aed28d0425a1d4a327", |
| "panel": 1 |
| } |
| ] |
| } |
| ] |
| }, |
| "egorenderer": { |
| "note": "该条目为自有合成渲染。下图是实际交付画面及配套 MANO 拟合。", |
| "figures": [ |
| { |
| "title": "EgoRenderer V005 的实际渲染与配套拟合", |
| "caption": "s1/airplane_lift · 151 · 双手 · 已交付合成数据 · forehead_independent_orbit", |
| "source_url": "https://yangzijing-nine-source-hand-data-browser.static.hf.space/hand_dataset_browser_20261013/index.html#egorenderer_v005", |
| "source_label": "团队 V005 图册", |
| "original_caption": "Existing paired beta+pose fit; not original GT. No extra wrist alignment. CPU projection checked", |
| "attribution": "EgoRenderer V005 · HOI-Cap 团队", |
| "images": [ |
| { |
| "src": "https://huggingface.co/spaces/yangzijing/nine-source-hand-data-browser/resolve/44fd810763cee24fe3d2f99c8502959e2c546c9f/hand_dataset_browser_20261013/media/egorenderer_v005_s1_airplane_lift_151_rgb.png", |
| "width": 1536, |
| "height": 864 |
| }, |
| { |
| "src": "https://huggingface.co/spaces/yangzijing/nine-source-hand-data-browser/resolve/44fd810763cee24fe3d2f99c8502959e2c546c9f/hand_dataset_browser_20261013/media/egorenderer_v005_s1_airplane_lift_151_fit.png", |
| "width": 1536, |
| "height": 864 |
| } |
| ] |
| } |
| ] |
| }, |
| "cocowb": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "作者发布的手部关键点标注示例", |
| "caption": "作者发布的手部关键点标注示例", |
| "source_url": "https://arxiv.org/html/2007.11858#S3.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Annotation examples for face/hand keypoints in COCO-WholeBody.", |
| "attribution": "Sheng Jin Lumin Xu Jin Xu Can Wang Wentao Liu Chen Qian Wanli Ouyang Ping Luo [.21cm] 1 The University of Hong Kong 2 SenseTime Research 3 The Chinese University of Hong Kong 4 The University of Sydney {jinsheng xulumin wangcan liuwentao qianchen}@sensetime.com wanli.ouyang@sydney.edu.au pluo@cs.hku.hk", |
| "images": [ |
| { |
| "src": "media/figures/cocowb_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2007.11858v1/images/hand_face_example.jpg", |
| "width": 4762, |
| "height": 420, |
| "sha256": "90d2440a8ef778c476d7751d9c482359448921abfa48921cac9b19abac062b4f", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "halpe": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "官方真实图像与全身、手部标注样例", |
| "caption": "Halpe 官方仓库发布的图像标注示例。", |
| "source_url": "https://github.com/Fang-Haoshu/Halpe-FullBody", |
| "source_label": "Halpe 官方数据说明", |
| "original_caption": "Halpe-FullBody official annotation examples.", |
| "attribution": "Halpe-FullBody · Fang-Haoshu et al.", |
| "images": [ |
| { |
| "src": "media/figures/halpe_annotation.webp", |
| "source_image_url": "https://github.com/Fang-Haoshu/Halpe-FullBody/raw/master/docs/1.jpg", |
| "width": 640, |
| "height": 480 |
| } |
| ] |
| }, |
| { |
| "title": "官方手部关键点定义", |
| "caption": "官方手部关键点定义", |
| "source_url": "https://github.com/Fang-Haoshu/Halpe-FullBody", |
| "source_label": "官方项目说明", |
| "original_caption": "官方手部关键点定义", |
| "attribution": "https://github.com/Fang-Haoshu/Halpe-FullBody", |
| "images": [ |
| { |
| "src": "media/figures/halpe_0_0.webp", |
| "source_image_url": "https://github.com/Fang-Haoshu/Halpe-FullBody/raw/master/docs/hand.jpg", |
| "width": 405, |
| "height": 512, |
| "sha256": "328e20e342afa70c31c3dbfb9a143681c4cf1c937ad842ea1b0ab4548c90278f", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "mpiinzsl": { |
| "note": "当前预览取 MPII / NZSL 的 manual_train 人工标注；下图为原作者展示的单图标注情况。", |
| "figures": [ |
| { |
| "title": "作者展示的遮挡与人工手部标注", |
| "caption": "作者展示的遮挡与人工手部标注", |
| "source_url": "https://arxiv.org/html/1704.07809#S1.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Hand annotation is difficult in single images because joints are often occluded due to (a) articulations of other parts of the hand, (b) a particular viewing angle, or (c) objects that the hand is grasping.", |
| "attribution": "Tomas Simon Hanbyul Joo Iain Matthews Yaser Sheikh Affiliation: Carnegie Mellon University Email: {tsimon,hanbyulj,iainm,yaser}@cs.cmu.edu", |
| "images": [ |
| { |
| "src": "media/figures/mpiinzsl_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/1704.07809v1/occlusion_types.png", |
| "width": 565, |
| "height": 159, |
| "sha256": "5a6ee4342acfc5136df6bf7c493944f2e631af24bb9c13ad2bbc9ff92a508ad2", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "ego4d": { |
| "note": "这里展示 Ego4D 原数据的采集内容；本站手部视频使用团队 P3R3 伪标签。", |
| "figures": [ |
| { |
| "title": "视频与 Matterport 三维扫描配对", |
| "caption": "视频与 Matterport 三维扫描配对", |
| "source_url": "https://arxiv.org/html/2110.07058#S3.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Figure 4 : Some videos (bottom) have coupled 3D meshes (top) from Matterport3D scanners, allowing one to relate the dynamic video to the static 3D environment (middle).", |
| "attribution": "Kristen Grauman Affiliation: Facebook AI Research (FAIR) Affiliation: University of Texas at Austin Andrew Westbury Affiliation: Facebook AI Research (FAIR) Eugene Byrne Zachary Chavis Antonino Furnari Rohit Girdhar Jackson Hamburger Hao Jiang Miao Liu Xingyu Liu Miguel Martin Tushar Nagarajan Affiliation: University of Texas at Austin Ilija Radosavovic Santhosh Kumar Ramakrishnan Affiliation: University of Texas at Austin Fiona Ryan Jayant Sharma Michael Wray Mengmeng Xu Eric Zhongcong Xu Chen Zhao Siddhant Bansal Affiliation: International Institute\nof Information Technology, Hyderabad Dhruv Batra Affiliation: Facebook AI Research (FAIR) Vincent Cartillier Affiliation: Facebook AI Research (FAIR) Affiliation: Georgia Tech Sean Crane Affiliation: Carnegie Mellon University Tien Do Affiliation: University of Minnesota Morrie Doulaty Affiliation: Facebook Akshay Erapalli Affiliation: Facebook Christoph Feichtenhofer Affiliation: Facebook AI Research (FAIR) Adriano Fragomeni Affiliation: University of Bristol Qichen Fu Affiliation: Carnegie Mellon University Abrham Gebreselasie Affiliation: Carnegie Mellon University Africa Cristina González Affiliation: Universidad de los Andes James Hillis Affiliation: Facebook Reality Labs Xuhua Huang Affiliation: Carnegie Mellon University Yifei Huang Affiliation: University of Tokyo Wenqi Jia Affiliation: Georgia Tech Weslie Khoo Affiliation: Indiana University Jáchym Kolář Affiliation: Facebook Satwik Kottur Affiliation: Facebook Anurag Kumar Affiliation: Facebook Reality Labs Federico Landini Affiliation: Facebook Chao Li Affiliation: Facebook Reality Labs Yanghao Li Affiliation: Facebook AI Research (FAIR) Zhenqiang Li Affiliation: University of Tokyo Karttikeya Mangalam Affiliation: Facebook AI Research (FAIR) Affiliation: UC Berkeley Raghava Modhugu Affiliation: International Institute\nof Information Technology, Hyderabad Jonathan Munro Affiliation: University of Bristol Tullie Murrell Affiliation: Facebook AI Research (FAIR) Takumi Nishiyasu Affiliation: University of Tokyo Will Price Affiliation: University of Bristol Paola Ruiz Puentes Affiliation: Universidad de los Andes Merey Ramazanova Affiliation: King Abdullah University of Science and Technology Leda Sari Affiliation: Facebook Reality Labs Kiran Somasundaram Affiliation: Facebook Reality Labs Audrey Southerland Affiliation: Georgia Tech Yusuke Sugano Affiliation: University of Tokyo Ruijie Tao Affiliation: National University of Singapore Minh Vo Affiliation: Facebook Reality Labs Yuchen Wang Affiliation: Indiana University Xindi Wu Affiliation: Carnegie Mellon University Takuma Yagi Affiliation: University of Tokyo Ziwei Zhao Affiliation: Indiana University Yunyi Zhu Affiliation: National University of Singapore Pablo Arbeláez David Crandall Dima Damen Giovanni Maria Farinella Christian Fuegen Bernard Ghanem Vamsi Krishna Ithapu C. V. Jawahar Hanbyul Joo Kris Kitani Haizhou Li Richard Newcombe Aude Oliva Hyun Soo Park James M. Rehg Yoichi Sato Jianbo Shi Mike Zheng Shou Antonio Torralba Lorenzo Torresani Affiliation: Dartmouth Mingfei Yan Jitendra Malik Affiliation: Facebook AI Research (FAIR) Affiliation: UC Berkeley Affiliation: University of Catania Affiliation: MIT Affiliation: University of Pennsylvania", |
| "images": [ |
| { |
| "src": "media/figures/ego4d_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2110.07058v3/matterport.png", |
| "width": 1456, |
| "height": 899, |
| "sha256": "194ff89bb88d83c9a37e7a3b626a452932a3c640f2b2b94a2d6ed328e0675c6e", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "epic": { |
| "note": "采集设备图来自 EPIC-KITCHENS 原论文；本站手部视频使用团队 P3R3 伪标签。", |
| "figures": [ |
| { |
| "title": "参与者佩戴的 GoPro 采集设备", |
| "caption": "参与者佩戴的 GoPro 采集设备", |
| "source_url": "https://arxiv.org/html/1804.02748#S3.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2: Head-mounted GoPro used in dataset recording", |
| "attribution": "Dima Damen 1 Hazel Doughty 1 Giovanni Maria Farinella 2 Sanja Fidler 3 Antonino Furnari 2 Evangelos Kazakos 1 Davide Moltisanti 1 Jonathan Munro 1 Toby Perrett 1 Will Price 1 Michael Wray 1 Affiliation: Uni. of Bristol, UK Uni. of Catania, Italy, Uni. of Toronto, Canada", |
| "images": [ |
| { |
| "src": "media/figures/epic_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/1804.02748v2/Figures-DM/setup_Davide.jpg", |
| "width": 877, |
| "height": 584, |
| "sha256": "a6604ece6e5fc124a5dd1e5d1196bd4c59c3dc7e121ca1fe3d521b473afeabb6", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "egodex": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "记录关节与真实操作样例", |
| "caption": "记录关节与真实操作样例", |
| "source_url": "https://arxiv.org/html/2505.11709#S3.F3", |
| "source_label": "论文图 3", |
| "original_caption": "Figure 3: Left : Joints captured by EgoDex. Right : Examples of dexterous manipulation behaviors. Tracked fingertips are highlighted in distinct colors and show 0.5 seconds of motion before the current frame. From left to right, top to bottom, the tasks are: unzipping a Ziploc bag, removing a book from a bookshelf, removing a screw from a fixture, folding a t-shirt, decluttering, opening a case, unscrewing a bottle cap, tying shoelaces, and washing a cup.", |
| "attribution": "Ryan Hoque * Peide Huang * David J. Yoon † † thanks: Equal contribution. Mouli Sivapurapu Jian Zhang Affiliation: Apple", |
| "images": [ |
| { |
| "src": "media/figures/egodex_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2505.11709v3/figs/skeleton_v2.png", |
| "width": 3240, |
| "height": 2475, |
| "sha256": "0e254fcad9a79b43ab7939d2e5dbe6f4ff78f6f0383b1a5e861afff98d5ab46f", |
| "panel": 0 |
| }, |
| { |
| "src": "media/figures/egodex_0_1.webp", |
| "source_image_url": "https://arxiv.org/html/2505.11709v3/figs/collage_edit.jpg", |
| "width": 5760, |
| "height": 3240, |
| "sha256": "14a638555e5e38167c556e8823f5527a7e534e08dabf90a917f4365b358e0162", |
| "panel": 1 |
| } |
| ] |
| } |
| ] |
| }, |
| "egopressure": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "7 台固定相机与头戴相机", |
| "caption": "7 台固定相机与头戴相机", |
| "source_url": "https://arxiv.org/html/2409.02224#S4.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Figure 3 : 7 static + 1 egocentric camera rig", |
| "attribution": "Yiming Zhao \nTaein Kwon \nPaul Streli \nMarc Pollefeys \nChristian Holz Affiliation: ETH Zürich Microsoft Spatial AI Lab, Zürich", |
| "images": [ |
| { |
| "src": "media/figures/egopressure_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2409.02224v2/camera_setup.png", |
| "width": 2024, |
| "height": 1122, |
| "sha256": "11efcb8ffd420464331856c1b1d76674fbb7c2bb29079f9f081c1ae477cea9f3", |
| "panel": 0 |
| }, |
| { |
| "src": "media/figures/egopressure_0_1.webp", |
| "source_image_url": "https://arxiv.org/html/2409.02224v2/IRtracking.png", |
| "width": 1998, |
| "height": 999, |
| "sha256": "dc7b60556bcd721321c75e1e4d8c94f71fe77444f0e892b93757b6b90b243923", |
| "panel": 1 |
| } |
| ] |
| }, |
| { |
| "title": "多视角 RGB-D 与压力联合标注流程", |
| "caption": "多视角 RGB-D 与压力联合标注流程", |
| "source_url": "https://arxiv.org/html/2409.02224#S3.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Method overview. The input for our annotation method consists of RGB-D images captured by 7 static Azure Kinect cameras and the pressure frame from a Sensel Morph touchpad. We leverage Segment-Anything [ 46 ] and HaMeR [ 67 ] to obtain initial hand poses and masks. We refine the initial hand pose and shape estimates through differentiable rasterization [ 9 ] optimization across all static camera views. Using an additional virtual orthogonal camera placed below the touchpad, we reproject the captured pressure frame onto the hand mesh by optimizing the pressure as a texture feature of the corresponding UV map, while ensuring contact between the touchpad and all contact vertices.", |
| "attribution": "Yiming Zhao \nTaein Kwon \nPaul Streli \nMarc Pollefeys \nChristian Holz Affiliation: ETH Zürich Microsoft Spatial AI Lab, Zürich", |
| "images": [ |
| { |
| "src": "media/figures/egopressure_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2409.02224v2/pipeline_anno.png", |
| "width": 2040, |
| "height": 612, |
| "sha256": "ec82d5e68ca56900faee915ec615fdb44e45784616b82ac319b3e3c9a7a27d17", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "egoemg": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "双腕 EMG、头戴相机与光捕采集系统", |
| "caption": "双腕 EMG、头戴相机与光捕采集系统", |
| "source_url": "https://arxiv.org/html/2605.05712#S3.F1", |
| "source_label": "论文图 1", |
| "original_caption": "Figure 1 : Data collection setup for EgoEMG . Bilateral EMG wristbands, head-mounted egocentric RGB, external ZED 2i RGB-D, and optical motion capture with hand markers for pose labels.", |
| "attribution": "Ziheng Xi Jiayi Yu Yitao Wang Yanbo Duan Jianjiang Feng Jie Zhou Affiliation: Department of Automation, Tsinghua University Affiliation: {xizh21, jy-yu23, wangyita23, duanyb24}@mails.tsinghua.edu.cn Affiliation: {jfeng, jzhou}@tsinghua.edu.cn", |
| "images": [ |
| { |
| "src": "media/figures/egoemg_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2605.05712v1/fig_capture_setup.png", |
| "width": 1144, |
| "height": 477, |
| "sha256": "714222b6a921ef15b359645a9dcbba0dbd31ad94862f794a0ac03257da98f754", |
| "panel": 0 |
| } |
| ] |
| }, |
| { |
| "title": "反光点到 MANO 的标注流程", |
| "caption": "反光点到 MANO 的标注流程", |
| "source_url": "https://arxiv.org/html/2605.05712#A2.F8", |
| "source_label": "论文图 8", |
| "original_caption": "Figure 8 : Architecture of markers2mano pipeline.", |
| "attribution": "Ziheng Xi Jiayi Yu Yitao Wang Yanbo Duan Jianjiang Feng Jie Zhou Affiliation: Department of Automation, Tsinghua University Affiliation: {xizh21, jy-yu23, wangyita23, duanyb24}@mails.tsinghua.edu.cn Affiliation: {jfeng, jzhou}@tsinghua.edu.cn", |
| "images": [ |
| { |
| "src": "media/figures/egoemg_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2605.05712v1/Graph_Transformer_Architecture.png", |
| "width": 2040, |
| "height": 600, |
| "sha256": "36454fd03c3bc9b6bdf94af06dffe89336ef7aca0a3c6f8deb6c64f6997438ea", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "hocap": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "HO-Cap 的真实采集装置", |
| "caption": "HO-Cap 的真实采集装置", |
| "source_url": "https://arxiv.org/html/2406.06843#S2.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Illustration of our data capture setup.", |
| "attribution": "Jikai Wang Affiliation: The University of Texas at Dallas Email: jikai.wang@utdallas.edu Qifan Zhang Affiliation: The University of Texas at Dallas Email: qifan.zhang@utdallas.edu Yu-Wei Chao Email: xguo@utdallas.edu Affiliation: NVIDIA Bowen Wen Email: yu.xiang@utdallas.edu Affiliation: NVIDIA Xiaohu Guo Affiliation: The University of Texas at Dallas Email: ychao@nvidia.com Yu Xiang Affiliation: The University of Texas at Dallas Email: bowenw@nvidia.com", |
| "images": [ |
| { |
| "src": "media/figures/hocap_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2406.06843v4/hardware.png", |
| "width": 1680, |
| "height": 464, |
| "sha256": "00cce88c9eb6a4043673286db25f8a10294cc333a8ff887d31fa21bfcc124921", |
| "panel": 0 |
| } |
| ] |
| }, |
| { |
| "title": "手与物体的姿态标注流程", |
| "caption": "手与物体的姿态标注流程", |
| "source_url": "https://arxiv.org/html/2406.06843#S4.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Figure 4 : Illustration of our pipeline for obtaining poses of hands and objects from multi-view RGB-D videos.", |
| "attribution": "Jikai Wang Affiliation: The University of Texas at Dallas Email: jikai.wang@utdallas.edu Qifan Zhang Affiliation: The University of Texas at Dallas Email: qifan.zhang@utdallas.edu Yu-Wei Chao Email: xguo@utdallas.edu Affiliation: NVIDIA Bowen Wen Email: yu.xiang@utdallas.edu Affiliation: NVIDIA Xiaohu Guo Affiliation: The University of Texas at Dallas Email: ychao@nvidia.com Yu Xiang Affiliation: The University of Texas at Dallas Email: bowenw@nvidia.com", |
| "images": [ |
| { |
| "src": "media/figures/hocap_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2406.06843v4/pipeline.png", |
| "width": 2048, |
| "height": 1725, |
| "sha256": "c085653a194701f225fe991b22331554dc6390339a2ee523ea52f5d13e001d54", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "taco": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "手、物体与接触的自动标注流程", |
| "caption": "手、物体与接触的自动标注流程", |
| "source_url": "https://arxiv.org/html/2401.08399#S3.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Automatic data annotating pipeline. The input consists of color frames from allocentric views, pre-scanned object models, and 3D positions of markers attached to object surfaces. We first separately localize 3D hand keypoints and obtain object poses, and then conduct contact-aware optimization to recover MANO [ 55 ] hand meshes. We finally segment hands and objects from images and automatically inpaint markers to acquire realistic object appearances.", |
| "attribution": "Yun Liu Affiliation: Tsinghua University Affiliation: Shanghai Artificial Intelligence Laboratory Affiliation: Shanghai Qi Zhi Institute Haolin Yang Affiliation: Beijing University of Posts and Telecommunications Xu Si Affiliation: Tsinghua University Ling Liu Affiliation: Beijing Institute of Technology Zipeng Li Affiliation: Tsinghua University Yuxiang Zhang Affiliation: Tsinghua University Yebin Liu Affiliation: Tsinghua University Li Yi Affiliation: Tsinghua University Affiliation: Shanghai Artificial Intelligence Laboratory Affiliation: Shanghai Qi Zhi Institute", |
| "images": [ |
| { |
| "src": "media/figures/taco_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2401.08399v2/imgs/pipeline_haolin.png", |
| "width": 5132, |
| "height": 1747, |
| "sha256": "27035148f5d0ede59034cf2b5f55874c6a41ccb0a7ea8a8272331cc8b589426b", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "aceda0": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "真实房间布局与相机布置", |
| "caption": "真实房间布局与相机布置", |
| "source_url": "https://arxiv.org/html/2607.28625#S3.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Overview of the Ambient Capture Engine. We design two complementary home scenes and capture at two spatial scales, with the equipment installed accordingly.", |
| "attribution": "1We introduce the Ambient Capture Engine (ACE), a capture system that transforms real home environments into recording studios. ACE observes household activity at two complementary scales, i.e., table-scale and room-scale. Via ACE, we capture and release ACE-Data-0 with synchronized multiple modalities and rich data annotations.", |
| "images": [ |
| { |
| "src": "media/figures/aceda0_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2607.28625v1/images/setup_ace-r.png", |
| "width": 1365, |
| "height": 1152, |
| "sha256": "3a8de57e80f04f3f4a4f5416ddecf16e566cbb22564f985554476f24d97e2f49", |
| "panel": 0 |
| }, |
| { |
| "src": "media/figures/aceda0_0_1.webp", |
| "source_image_url": "https://arxiv.org/html/2607.28625v1/setup_ace-t.png", |
| "width": 1183, |
| "height": 882, |
| "sha256": "f538a521d7ac250f019d43f3b9fde157d1ac7188077e0140c981d2a328c034f5", |
| "panel": 1 |
| } |
| ] |
| }, |
| { |
| "title": "ACE 多模态采集与标注流程", |
| "caption": "ACE 多模态采集与标注流程", |
| "source_url": "https://arxiv.org/html/2607.28625#S4.F3", |
| "source_label": "论文图 3", |
| "original_caption": "Figure 3 : Overall workflow of ACE. Each participant wears the motion-capture suit before the multi-modal recording. Synchronization is conducted after data export to align the timeline of each modality. Finally, we annotate the collected data with rich and high-quality labels derived from the captured ground truth.", |
| "attribution": "1We introduce the Ambient Capture Engine (ACE), a capture system that transforms real home environments into recording studios. ACE observes household activity at two complementary scales, i.e., table-scale and room-scale. Via ACE, we capture and release ACE-Data-0 with synchronized multiple modalities and rich data annotations.", |
| "images": [ |
| { |
| "src": "media/figures/aceda0_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2607.28625v1/data_procedure.png", |
| "width": 1607, |
| "height": 880, |
| "sha256": "5c085debff49030635e8e1262b65f0412ac0a914ea9331025271762042c8bad2", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "xperience": { |
| "note": "当前取得的是作者发布的数据查看器截图；尚未取得可核对的采集硬件图。", |
| "figures": [ |
| { |
| "title": "作者数据查看器中的相机、深度与动作", |
| "caption": "作者数据查看器中的相机、深度与动作", |
| "source_url": "https://github.com/Ropedia/HOMIE-toolkit", |
| "source_label": "官方项目说明", |
| "original_caption": "Rerun visualization", |
| "attribution": "https://github.com/Ropedia/HOMIE-toolkit", |
| "images": [ |
| { |
| "src": "media/figures/xperience_0_0.webp", |
| "source_image_url": "https://github.com/Ropedia/HOMIE-toolkit/raw/main/assets/rerun.png", |
| "width": 2700, |
| "height": 1594, |
| "sha256": "ccdf9c229651485c50c986f195ffa206a2cc43f0f7e23d7db42fbf302ecfeeee", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "feel": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "六点力传感手套的真实硬件", |
| "caption": "六点力传感手套的真实硬件", |
| "source_url": "https://arxiv.org/html/2603.15847#S3.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Figure 4 : Force-sensing glove hardware. Leftmost: Palm-facing views of the left and right gloves, showing the six piezoresistive sensors ( red ) mounted at each fingertip and across the palm, the Arduino microcontroller case ( blue ) mounted at the wrist, and the on/off switch ( green ). Middle and right: Side and in-use views demonstrating the glove’s low-profile form factor during natural object manipulation.", |
| "attribution": "Eadom Dessalene Botao He Michael Maynord Yonatan Tussa Pavan Mantripragada Yianni Karabatis Nirupam Roy Yiannis Aloimonos Affiliation: University of Maryland, College Park, USA E-mail Email: edessale@umd.edu", |
| "images": [ |
| { |
| "src": "media/figures/feel_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2603.15847v1/Fig_5.png", |
| "width": 6413, |
| "height": 2420, |
| "sha256": "c01a5612f8d420e6b3e2640b6bc1a1ed71210012895f107e36ab954d08d954c3", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "openaoe": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "手机采集与结构化数据处理流程", |
| "caption": "手机采集与结构化数据处理流程", |
| "source_url": "https://arxiv.org/html/2607.14183#S0.F1", |
| "source_label": "论文图 1", |
| "original_caption": "Figure 1: Overview of Open-AoE. Open-AoE provides independent modules for smartphone capture, structured data processing, and downstream robot learning, supporting 2,000 hours of data across 500+ contributors, 400+ device types, 400+ scenes, and 8,000+ tasks.", |
| "attribution": "Ant Digital Technologies Ant Group Affiliation: GitHub | | Hugging Face | | ModelScope", |
| "images": [ |
| { |
| "src": "media/figures/openaoe_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2607.14183v2/figure1_v5.png", |
| "width": 1675, |
| "height": 887, |
| "sha256": "82da068713755d07269bd464fecf5a95d98c41ba9ad3bcab2ce190a4d0a83109", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "genhumanego": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "官方视频到结构化动作的处理流程", |
| "caption": "官方视频到结构化动作的处理流程", |
| "source_url": "https://www.gen-robot.org/blog/gen-human-ego", |
| "source_label": "官方项目说明", |
| "original_caption": "Gen-HumanEgo processing pipeline from raw egocentric video to structured motion representations", |
| "attribution": "https://www.gen-robot.org/blog/gen-human-ego", |
| "images": [ |
| { |
| "src": "media/figures/genhumanego_0_0.webp", |
| "width": 3200, |
| "height": 1080, |
| "source_image_url": "https://www.gen-robot.org/api/static/s3_manager?path=blog%2FGen-HumanEgo2%2Fgen-human-ego-raw-to-smr.png" |
| } |
| ] |
| } |
| ] |
| }, |
| "egolive": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "JoyEgoCam 头戴双目采集系统", |
| "caption": "JoyEgoCam 头戴双目采集系统", |
| "source_url": "https://arxiv.org/html/2604.23570#S2.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Human Data Capture System The system leverages JoyEgoCam which is a custom-designed head-mounted device for acquiring human behavior data in real-world environments. It features stereo RGB cameras that provide a wide field-of-view and an integrated IMU with 200Hz measurements.", |
| "attribution": "EgoLive: A Large-Scale Egocentric Dataset from Real-World Human Tasks", |
| "images": [ |
| { |
| "src": "media/figures/egolive_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2604.23570v1/figs/HumanDataCaptureSystem7.png", |
| "width": 1260, |
| "height": 882, |
| "sha256": "52dd49afbbe27e6ed7dcfdd4148679fcc3a1187f1d2d2cea3b996a6e81c3944c", |
| "panel": 0 |
| } |
| ] |
| }, |
| { |
| "title": "双目视频自动标注流程", |
| "caption": "双目视频自动标注流程", |
| "source_url": "https://arxiv.org/html/2604.23570#S3.F3", |
| "source_label": "论文图 3", |
| "original_caption": "Figure 3 : Overview of the automated annotation pipeline for egocentric binocular videos.\n(a) The raw egocentric binocular video is first undistorted, with left/right view separation and camera intrinsic calibration.\n(b) Hand reconstruction is conducted, integrating semantic gesture recognition and hand-object segmentation to capture interaction dynamics.\n(c) Dense depth reconstruction and camera localization are performed to recover the scene structure and estimate the camera trajectory based on the rectified video and sensor cues.\n(d) In parallel, sub-task segmentation and instruction captioning are employed to generate temporal textual description.", |
| "attribution": "EgoLive: A Large-Scale Egocentric Dataset from Real-World Human Tasks", |
| "images": [ |
| { |
| "src": "media/figures/egolive_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2604.23570v1/figs/liuchengkuangtu4.png", |
| "width": 1816, |
| "height": 1106, |
| "sha256": "af40ea531dac7bed1c60e3247617703097b520e3ae00f9e9dc085c3b42352796", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "egoverse": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "Aria、手机与行业采集设备", |
| "caption": "Aria、手机与行业采集设备", |
| "source_url": "https://arxiv.org/html/2604.07607#S3.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Fig. 2: Human Data Capture Setup. (Left) EgoVerse is captured through a variety of hardware systems, including Project Aria glasses (academic labs), a phone-based capture system (accessible by everyone), and custom setups by industry partners. (Right) Regardless of sources, human data is processed into a unified format that contains at minimum egocentric videos, hand keypoints, and camera poses.", |
| "attribution": "Ryan Punamiya Simar Kareer Zeyi Liu Affiliation: Stanford University Josh Citron Affiliation: Stanford University Ri-Zhao Qiu Affiliation: University of California San Diego Xiongyi Cai Affiliation: University of California San Diego Alexey Gavryushin Affiliation: ETH Zürich Jiaqi Chen Affiliation: ETH Zürich Davide Liconti\nLawrence Y. Zhu Affiliation: Georgia Institute of Technology Affiliation: ETH Zürich Patcharapong Aphiwetsa Affiliation: Georgia Institute of Technology Baoyu Li Affiliation: Georgia Institute of Technology Aniketh Cheluva Affiliation: Georgia Institute of Technology Pranav Kuppili Affiliation: Georgia Institute of Technology Yangcen Liu Affiliation: Georgia Institute of Technology Dhruv Patel Affiliation: Georgia Institute of Technology Aidan Gao Affiliation: Georgia Institute of Technology Hye-Young Chung Affiliation: Georgia Institute of Technology Ryan Co Affiliation: Georgia Institute of Technology Renee Zbizika Affiliation: Stanford University Jeff Liu Affiliation: Stanford University Xiaomeng Xu Affiliation: Stanford University Haoyu Xiong Affiliation: MIT CSAIL Geng Chen Affiliation: University of California San Diego Sebastiano Oliani Affiliation: ETH Zürich Wenkai Xuan Affiliation: ETH Zürich Chenyu Yang Affiliation: ETH Zürich Xi Wang Affiliation: ETH Zürich James Fort Affiliation: Meta Reality Labs Research Richard Newcombe Affiliation: Meta Reality Labs Research Josh Gao Affiliation: Mecka AI Jason Chong Affiliation: Mecka AI Garrett Matsuda Affiliation: Scale AI Aseem Doriwala Affiliation: Scale AI Marc Pollefeys Affiliation: ETH Zürich Robert Katzschmann Affiliation: ETH Zürich Xiaolong Wang Affiliation: University of California San Diego Shuran Song Affiliation: Stanford University Judy Hoffman Affiliation: Georgia Institute of Technology Danfei Xu Affiliation: Georgia Institute of Technology", |
| "images": [ |
| { |
| "src": "media/figures/egoverse_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2604.07607v2/figures/capture_setup.jpg", |
| "width": 1758, |
| "height": 964, |
| "sha256": "cdd961da050c6919ff4a6229affca6c500a18167415264ea5b186325977df78b", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "ipnhands": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "IPN HandS 的真实标注工具界面", |
| "caption": "作者发布的标注工具：手别、关键点、时间线和手工修正。", |
| "source_url": "https://www.mdpi.com/2076-3417/15/11/6321", |
| "source_label": "Applied Sciences 2025 · Figure 2", |
| "original_caption": "作者发布的标注工具：手别、关键点、时间线和手工修正。", |
| "attribution": "Gibran Benitez-Garcia, Jesus Olivares-Mercado, Gabriel Sanchez-Perez, Hiroki Takahashi", |
| "images": [ |
| { |
| "src": "media/figures/ipnhands_2.webp", |
| "source_image_url": "https://mdpi-res.com/applsci/applsci-15-06321/article_deploy/html/images/applsci-15-06321-g002.png", |
| "width": 3298, |
| "height": 2689 |
| } |
| ] |
| }, |
| { |
| "title": "IPN HandS 的手势类别与骨架样例", |
| "caption": "原文展示的手势类别及随时间变化的标注帧。", |
| "source_url": "https://www.mdpi.com/2076-3417/15/11/6321", |
| "source_label": "Applied Sciences 2025 · Figure 6", |
| "original_caption": "原文展示的手势类别及随时间变化的标注帧。", |
| "attribution": "Gibran Benitez-Garcia, Jesus Olivares-Mercado, Gabriel Sanchez-Perez, Hiroki Takahashi", |
| "images": [ |
| { |
| "src": "media/figures/ipnhands_6.webp", |
| "source_image_url": "https://mdpi-res.com/applsci/applsci-15-06321/article_deploy/html/images/applsci-15-06321-g006.png", |
| "width": 3947, |
| "height": 1503 |
| } |
| ] |
| } |
| ] |
| }, |
| "hograspnet": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "抓握类别与真实相机布置", |
| "caption": "抓握类别与真实相机布置", |
| "source_url": "https://arxiv.org/html/2409.04033#S3.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Figure 4 : (left) Per-object taxonomy examples (right) System setup. The full list is shown in the supplementary.", |
| "attribution": "Woojin Cho Jihyun Lee Minjae Yi Minje Kim Taeyun Woo Donghwan Kim Taewook Ha Hyokeun Lee Je-Hwan Ryu Woontack Woo Tae-Kyun Kim Affiliation: KAIST, Imperial College London, Kwangwoon University, Surromind", |
| "images": [ |
| { |
| "src": "media/figures/hograspnet_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2409.04033v1/figure/object_taxonomy_example.PNG", |
| "width": 1356, |
| "height": 948, |
| "sha256": "f4f36b78cc1ef42ee1de894a664d447ab9fe792ab122b4c8169098781cb621fc", |
| "panel": 0 |
| }, |
| { |
| "src": "media/figures/hograspnet_0_1.webp", |
| "source_image_url": "https://arxiv.org/html/2409.04033v1/figure/HardwareSetup.png", |
| "width": 1195, |
| "height": 684, |
| "sha256": "3242076dab6857a361761a3a419fa50b0e7613972756a98866c2ad2272b67e03", |
| "panel": 1 |
| } |
| ] |
| }, |
| { |
| "title": "MANO 与物体的标注流程", |
| "caption": "MANO 与物体的标注流程", |
| "source_url": "https://arxiv.org/html/2409.04033#S3.F6", |
| "source_label": "论文图 6", |
| "original_caption": "Figure 6 : MANO [ 68 ] and object annotation pipeline (Section 3.5 ).", |
| "attribution": "Woojin Cho Jihyun Lee Minjae Yi Minje Kim Taeyun Woo Donghwan Kim Taewook Ha Hyokeun Lee Je-Hwan Ryu Woontack Woo Tae-Kyun Kim Affiliation: KAIST, Imperial College London, Kwangwoon University, Surromind", |
| "images": [ |
| { |
| "src": "media/figures/hograspnet_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2409.04033v1/figure/AnnotationDiagram_update.png", |
| "width": 1308, |
| "height": 189, |
| "sha256": "a428e881d21aa9a524b99af7e332fe02833ead44121736b4bfc88c8ac37bdd97", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "palm": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "3dMD 多相机手部扫描系统", |
| "caption": "3dMD 多相机手部扫描系统", |
| "source_url": "https://arxiv.org/html/2511.05403#S3.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Capture setup . Our 3dMD setup with 7 7 RGB cameras.", |
| "attribution": "Zicong Fan Affiliation: Meta Reality Labs ETH Zürich Max Planck Institute for Intelligent Systems, Tübingen Edoardo Remelli David Dimond Fadime Sener Liuhao Ge Bugra Tekin Cem Keskin Shreyas Hampali Affiliation:", |
| "images": [ |
| { |
| "src": "media/figures/palm_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2511.05403v3/1_figures/dataset_setup.jpg", |
| "width": 1096, |
| "height": 1097, |
| "sha256": "c242c782a2c8aa06fe19551ff329f393c907152b4ad3f930744d7d700c9d5e1f", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "egoexo4d": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "Aria 第一视角采集设备", |
| "caption": "Aria 第一视角采集设备", |
| "source_url": "https://arxiv.org/html/2311.18259#S3.F3", |
| "source_label": "论文图 3", |
| "original_caption": "Figure 3 : The Aria device used for egocentric recordings.", |
| "attribution": "Kristen Grauman Affiliation: FAIR, Meta Affiliation: University of Texas at Austin Andrew Westbury Affiliation: FAIR, Meta Lorenzo Torresani Affiliation: FAIR, Meta Kris Kitani Affiliation: FAIR, Meta Affiliation: Carnegie Mellon University Jitendra Malik Affiliation: FAIR, Meta Affiliation: University of California, Berkeley Triantafyllos Afouras ∗ Affiliation: FAIR, Meta Kumar Ashutosh ∗ Affiliation: FAIR, Meta Affiliation: University of Texas at Austin Vijay Baiyya ∗ Affiliation: Project Aria, Meta Siddhant Bansal ∗ Affiliation: University of Bristol Affiliation: IIIT, Hyderabad Bikram Boote ∗ Affiliation: University of Illinois, Urbana Champaign Eugene Byrne ∗ Affiliation: FAIR, Meta Affiliation: Carnegie Mellon University Zach Chavis ∗ Affiliation: University of Minnesota Joya Chen ∗ Affiliation: National University of Singapore Feng Cheng ∗ Affiliation: FAIR, Meta Fu-Jen Chu ∗ Affiliation: FAIR, Meta Sean Crane ∗ Affiliation: Carnegie Mellon University Avijit Dasgupta ∗ Affiliation: IIIT, Hyderabad Jing Dong ∗ Affiliation: Project Aria, Meta Maria Escobar ∗ Affiliation: Universidad de los Andes Cristhian Forigua ∗ Affiliation: Universidad de los Andes Abrham Gebreselasie ∗ Affiliation: Carnegie Mellon University Sanjay Haresh ∗ Affiliation: Simon Fraser University Jing Huang ∗ Affiliation: FAIR, Meta Md Mohaiminul Islam ∗ Affiliation: University of North Carolina, Chapel Hill Suyog Jain ∗ Affiliation: FAIR, Meta Rawal Khirodkar ∗ Affiliation: Carnegie Mellon University Devansh Kukreja ∗ Affiliation: FAIR, Meta Kevin J Liang ∗ Affiliation: FAIR, Meta Jia-Wei Liu ∗ Affiliation: National University of Singapore Sagnik Majumder ∗ Affiliation: FAIR, Meta Affiliation: University of Texas at Austin Yongsen Mao ∗ Affiliation: Simon Fraser University Miguel Martin ∗ Affiliation: FAIR, Meta Effrosyni Mavroudi ∗ Affiliation: FAIR, Meta Tushar Nagarajan ∗ Affiliation: FAIR, Meta Francesco Ragusa ∗ Affiliation: University of Catania Santhosh Kumar Ramakrishnan ∗ Affiliation: University of Texas at Austin Luigi Seminara ∗ Affiliation: University of Catania Arjun Somayazulu ∗ Affiliation: University of Texas at Austin Yale Song ∗ Affiliation: FAIR, Meta Shan Su ∗ Affiliation: University of Pennsylvania Zihui Xue ∗ Affiliation: FAIR, Meta Affiliation: University of Texas at Austin Edward Zhang ∗ Affiliation: University of Pennsylvania Jinxu Zhang ∗ Affiliation: University of Pennsylvania Angela Castillo Affiliation: Universidad de los Andes Changan Chen Affiliation: University of Texas at Austin Xinzhu Fu Affiliation: National University of Singapore Ryosuke Furuta Affiliation: University of Tokyo Cristina González Affiliation: Universidad de los Andes Prince Gupta Affiliation: Project Aria, Meta Jiabo Hu Affiliation: Meta Yifei Huang Affiliation: University of Tokyo Yiming Huang Affiliation: University of Pennsylvania Weslie Khoo Affiliation: Indiana University Anush Kumar Affiliation: University of Minnesota Robert Kuo Affiliation: Meta Sach Lakhavani Affiliation: Project Aria, Meta Miao Liu Affiliation: Meta Mi Luo Affiliation: University of Texas at Austin Zhengyi Luo Affiliation: Carnegie Mellon University Brighid Meredith Affiliation: Meta Austin Miller Affiliation: Meta Oluwatumininu Oguntola Affiliation: University of North Carolina, Chapel Hill Xiaqing Pan Affiliation: Project Aria, Meta Penny Peng Affiliation: Meta Shraman Pramanick Affiliation: Johns Hopkins University Merey Ramazanova Affiliation: King Abdullah University of Science and Technology Fiona Ryan Affiliation: Georgia Tech Wei Shan Affiliation: University of North Carolina, Chapel Hill Kiran Somasundaram Affiliation: Project Aria, Meta Chenan Song Affiliation: National University of Singapore Audrey Southerland Affiliation: Georgia Tech Masatoshi Tateno Affiliation: University of Tokyo Huiyu Wang Affiliation: FAIR, Meta Yuchen Wang Affiliation: Indiana University Takuma Yagi Affiliation: University of Tokyo Mingfei Yan Affiliation: Project Aria, Meta Xitong Yang Affiliation: FAIR, Meta Zecheng Yu Affiliation: University of Tokyo Shengxin Cindy Zha Affiliation: Meta Chen Zhao Affiliation: King Abdullah University of Science and Technology Ziwei Zhao Affiliation: Indiana University Zhifan Zhu Affiliation: University of Bristol Jeff Zhuo Affiliation: University of North Carolina, Chapel Hill Pablo Arbeláez † Affiliation: Universidad de los Andes Gedas Bertasius † Affiliation: University of North Carolina, Chapel Hill David Crandall † Affiliation: Indiana University Dima Damen † Affiliation: University of Bristol Jakob Engel † Affiliation: Project Aria, Meta Giovanni Maria Farinella † Affiliation: University of Catania Antonino Furnari † Affiliation: University of Catania Bernard Ghanem † Affiliation: King Abdullah University of Science and Technology Judy Hoffman † Affiliation: Georgia Tech C. V. Jawahar † Affiliation: IIIT, Hyderabad Richard Newcombe † Affiliation: Project Aria, Meta Hyun Soo Park † Affiliation: University of Minnesota James M. Rehg † Affiliation: University of Illinois, Urbana Champaign Yoichi Sato † Affiliation: University of Tokyo Manolis Savva † Affiliation: Simon Fraser University Jianbo Shi † Affiliation: University of Pennsylvania Mike Zheng Shou † Affiliation: National University of Singapore Michael Wray † Affiliation: University of Bristol", |
| "images": [ |
| { |
| "src": "media/figures/egoexo4d_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2311.18259v4/figs/aria/aria.png", |
| "width": 655, |
| "height": 248, |
| "sha256": "d04ba99bfff6f60aa243b09add1f6453420c33135ddc524c4c58760e93bc18a1", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "gigahands": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "多视角手部跟踪与标注流程", |
| "caption": "多视角手部跟踪与标注流程", |
| "source_url": "https://arxiv.org/html/2412.04244#as1_S7.F5", |
| "source_label": "论文图 5", |
| "original_caption": "Figure 5 : Hand Motion Tracking Pipeline.", |
| "attribution": "Rao Fu Dingxi Zhang Alex Jiang Affiliation: Brown University Wanjia Fu Affiliation: Brown University Austin Funk Affiliation: Brown University Daniel Ritchie Affiliation: Brown University Srinath Sridhar * Equal contribution Corresponding author Affiliation: ETH Zurich https://ivl.cs.brown.edu/research/gigahands.html", |
| "images": [ |
| { |
| "src": "media/figures/gigahands_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2412.04244v3/hand_tracking.png", |
| "width": 1800, |
| "height": 450, |
| "sha256": "49151e95799601ea30ad9db7aeede15ce467b1e51f3071d31bcd60a73d59df5e", |
| "panel": 0 |
| } |
| ] |
| }, |
| { |
| "title": "作者的人工标注界面", |
| "caption": "作者的人工标注界面", |
| "source_url": "https://arxiv.org/html/2412.04244#as1_S9.F7", |
| "source_label": "论文图 7", |
| "original_caption": "Figure 7 : Annotation Interface.", |
| "attribution": "Rao Fu Dingxi Zhang Alex Jiang Affiliation: Brown University Wanjia Fu Affiliation: Brown University Austin Funk Affiliation: Brown University Daniel Ritchie Affiliation: Brown University Srinath Sridhar * Equal contribution Corresponding author Affiliation: ETH Zurich https://ivl.cs.brown.edu/research/gigahands.html", |
| "images": [ |
| { |
| "src": "media/figures/gigahands_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2412.04244v3/supp_img/annotation_interface.png", |
| "width": 2154, |
| "height": 1362, |
| "sha256": "5247c69f379664c1366d5f653f18df7f395baa03d164468f4071ea72efb41116", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "oakink2": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "12 台光捕相机与 4 台 RGB 相机的真实采集平台", |
| "caption": "原论文第 4 页图 3：蓝色标出光捕相机，红色标出 RGB 相机。", |
| "source_url": "https://openaccess.thecvf.com/content/CVPR2024/papers/Zhan_OAKINK2_A_Dataset_of_Bimanual_Hands-Object_Manipulation_in_Complex_Task_CVPR_2024_paper.pdf#page=4", |
| "source_label": "CVPR 2024 · Figure 3", |
| "original_caption": "Figure 3. Capture platform. 12 MoCap cameras are circled in blue and 4 RGB cameras in red.", |
| "attribution": "OAKINK2 · Qihang Zhan et al., CVPR 2024", |
| "images": [ |
| { |
| "src": "media/figures/oakink2_capture.webp", |
| "width": 726, |
| "height": 486 |
| } |
| ] |
| } |
| ] |
| }, |
| "hint": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "HInt 数据来源与关键点标注示例", |
| "caption": "HInt 数据来源与关键点标注示例", |
| "source_url": "https://arxiv.org/html/2312.05251#S2.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2: Dataset and Architecture. (Top) Hand crops with keypoint annotations from our HInt dataset of annotations for different image sources, Hands23 [ 9 ] , Epic-Kitchens [ 13 , 12 ] , and Ego4D [ 18 ] . We provide location annotations for 21 hand keypoints as well as the “occlusion” label for each joint.\nOccluded keypoints are marked using solid dot filled with black while non-occluded ones are filled with white.\nThe pie chart shows the distribution and statistics of our dataset. (Bottom) The architecture for HaMeR follows a fully transformer-based design.\nWe use a large scale ViT backbone [ 14 ] followed by a transformer decoder to regress the parameters of the hand.", |
| "attribution": "Georgios Pavlakos Dandan Shan Ilija Radosavovic Angjoo Kanazawa David Fouhey Jitendra Malik UC Berkeley University of Michigan New York University", |
| "images": [ |
| { |
| "src": "media/figures/hint_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2312.05251v1/dataset_architecture.png", |
| "width": 694, |
| "height": 247, |
| "sha256": "bc09248dccb9958aefb5219a6d3958c7fb0fdd3dff33f5207183851bc72f5531", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "blurhand": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "从连续清晰帧生成运动模糊图像", |
| "caption": "从连续清晰帧生成运动模糊图像", |
| "source_url": "https://arxiv.org/pdf/2303.15417#page=3", |
| "source_label": "CVPR 2023 · Figure 2", |
| "original_caption": "Figure 2 : Pipeline for constructing our BlurHand dataset. We synthesize the blurry hand from five sequential sharp hand frames by adopting video interpolation and averaging them.", |
| "attribution": "Yeonguk Oh JoonKyu Park Jaeha Kim Gyeongsik Moon Affiliation: Meta Reality Labs Research {namepllet, jkpark0825}@snu.ac.kr, jhkim97s2@gmail.com, mks0601@meta.com, kyoungmu@snu.ac.kr Kyoung Mu Lee Affiliation: Dept. of ECE&ASRI Affiliation: IPAI, Seoul National University, Korea", |
| "images": [ |
| { |
| "src": "media/figures/blurhand_0_0.webp", |
| "width": 723, |
| "height": 852 |
| } |
| ] |
| } |
| ] |
| }, |
| "dartset": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "DART 合成数据生成界面", |
| "caption": "DART 合成数据生成界面", |
| "source_url": "https://arxiv.org/html/2210.07650#S3.F3", |
| "source_label": "论文图 3", |
| "original_caption": "Figure 3: DART GUI for synthetic data generation. It supports adding textures, deforming skeleton, changing illuminations and backgrounds, and exporting the MANO poses.", |
| "attribution": "Daiheng Gao Email: daiheng.gdh@alibaba-inc.com Yuliang Xiu Email: yunzong.wf@alibaba-inc.com Kailin Li Email: funtian.zp@alibaba-inc.com Lixin Yang Email: zhangbang.zb@alibaba-inc.com Feng Wang Affiliation: Alibaba XR Lab Email: yuliang.xiu@tuebingen.mpg.de Peng Zhang Affiliation: Alibaba XR Lab Email: kailinli@sjtu.edu.cn Bang Zhang Affiliation: Alibaba XR Lab Email: siriusyang@sjtu.edu.cn Cewu Lu Affiliation: Shanghai Jiao Tong University Email: lucewu@sjtu.edu.cn Ping Tan Affiliation: Simon Fraser University Email: pingtan@sfu.ca Affiliation: Max Planck Institute for Intelligent Systems", |
| "images": [ |
| { |
| "src": "media/figures/dartset_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2210.07650v1/images/gui.png", |
| "width": 1754, |
| "height": 985, |
| "sha256": "64b1a913adc5228e894db46adc1a551fb3c536924289229ac3fd56bf3e35405f", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "humbi": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "多相机系统的真实布局设计", |
| "caption": "多相机系统的真实布局设计", |
| "source_url": "https://arxiv.org/html/2110.00119#S2.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Fig. 4: Re-configurable dodecagon design and its dimension of the multi-camera system.", |
| "attribution": "Jae Shin Yoon Zhixuan Yu Jaesik Park Hyun Soo Park † † thanks: J. S. Yoon, Z. Yu, and H. S. Park are with University of Minnesota. E-mail: {jsyoon, yu000064, hspark}@umn.edu\nJ. Park is with POSTECH. E-mail: jaesik.park@postech.ac.kr", |
| "images": [ |
| { |
| "src": "media/figures/humbi_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2110.00119v2/stage.png", |
| "width": 2038, |
| "height": 421, |
| "sha256": "8f317ae930889a206a1c113e0a0a4e264afb7fce497eff1e236c4f84df125767", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "ourcapture": { |
| "note": "现场照片与装置图待补充。", |
| "figures": [] |
| }, |
| "assemblyhands": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "多视角手部标注与数据集构建流程", |
| "caption": "多视角手部标注与数据集构建流程", |
| "source_url": "https://arxiv.org/html/2304.12301#S1.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Construction of AssemblyHands dataset and a benchmark task for egocentric 3D hand pose estimation. We first use manual annotations and an automatic annotation network (MVExoNet) to generate accurate 3D hand poses for multi-view images sampled from the Assembly101 dataset [ 28 ] .\nThese annotations are used to train a single-view 3D hand pose estimation network (SVEgoNet) from egocentric images.\nFinally, the predicted hand poses are evaluated by the action classification task.", |
| "attribution": "Takehiko Ohkawa 1 Affiliation: 1 Meta Reality Labs 2 The University of Tokyo 2] * * footnotemark: * Kun He 1 Fadime Sener 1 Tomas Hodan 1 Luan Tran 1 Cem Keskin 1 Affiliation: https://assemblyhands.github.io", |
| "images": [ |
| { |
| "src": "media/figures/assemblyhands_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2304.12301v1/pipeline.png", |
| "width": 2028, |
| "height": 729, |
| "sha256": "ffe70b1c4c03720bc0e42420b0cadd9b0b4b28556643e8a66455c2f86df086ff", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "show3d": { |
| "note": "", |
| "figures": [ |
| { |
| "title": "移动式半球相机采集背包", |
| "caption": "移动式半球相机采集背包", |
| "source_url": "https://arxiv.org/html/2603.28760#S1.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2 : Our mobile multi-camera capture rig. Left: Annotated hardware layout showing five MoCap cameras (red), and eight exocentric monochrome cameras mounted in a half-dome configuration (green), and two egocentric monochrome cameras on the Meta Quest 3 headset (blue).\nThe MoCap cameras are only used for headset pose tracking, by tracking optical markers rigidly attached to the headset.\nThe ten total exocentric and egocentric monochrome cameras are used for marker-free annotation of 3D hand and object poses.\nRight: The rig in use during in-the-wild capture sessions, demonstrating its lightweight (about eight kilograms) and wearable design that allows natural interaction while maintaining stable, synchronized multi-view coverage under mostly unconstrained motion.", |
| "attribution": "Patrick Rim 1,2 Kevin Harris 1 Braden Copple 1 Shangchen Han 1 Xu Xie 1 Ivan Shugurov 1 Sizhe An 1 He Wen 1 Alex Wong 2 Tomas Hodan 1 Kun He 1 1 Meta Reality Labs 2 Yale University † † thanks:", |
| "images": [ |
| { |
| "src": "media/figures/show3d_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2603.28760v1/SHOW3D_Figure_Rig.png", |
| "width": 2048, |
| "height": 884, |
| "sha256": "959a48035432610cd34c955c35adb6e82ab44271237ae1200c06a1429b150709", |
| "panel": 0 |
| } |
| ] |
| }, |
| { |
| "title": "ego-exo 手与物体标注流程", |
| "caption": "ego-exo 手与物体标注流程", |
| "source_url": "https://arxiv.org/html/2603.28760#S1.F3", |
| "source_label": "论文图 3", |
| "original_caption": "Figure 3 : Our ego-exo pipeline for 3D hand and object pose annotation. (a) Multi-view fisheye images from our ego and exo cameras.\n(b) We detect 3D hand keypoints by fusing predictions from Sapiens [ 14 ] and InterNet [ 22 ] , and fit personalized hand mesh via Inverse Kinematics.\n(c) CAD-based 3D object pose estimation using CNOS [ 25 ] , FoundPose [ 28 ] , and GoTrack [ 24 ] .\n(d) The resulting 3D ground-truth annotations are projected back into ego cameras, and can be used to train egocentric vision models.", |
| "attribution": "Patrick Rim 1,2 Kevin Harris 1 Braden Copple 1 Shangchen Han 1 Xu Xie 1 Ivan Shugurov 1 Sizhe An 1 He Wen 1 Alex Wong 2 Tomas Hodan 1 Kun He 1 1 Meta Reality Labs 2 Yale University † † thanks:", |
| "images": [ |
| { |
| "src": "media/figures/show3d_1_0.webp", |
| "source_image_url": "https://arxiv.org/html/2603.28760v1/Figure_Pipeline.png", |
| "width": 2027, |
| "height": 740, |
| "sha256": "4463f2cadf6d3cf6de46ac405755c939ca2dfbd23e016fccaf72c071462d31c7", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "hiphi": { |
| "note": "该数据集发布动作而无配套 RGB；这里展示作者的数据构建流程。", |
| "figures": [ |
| { |
| "title": "动作种子与数据构建流程", |
| "caption": "动作种子与数据构建流程", |
| "source_url": "https://arxiv.org/html/2608.16222#S3.F2", |
| "source_label": "论文图 2", |
| "original_caption": "Figure 2: From FrameNet to HiPHI data construction. Motion-relevant frames and lexical units define motion seeds, which are expanded through controlled factors such as speed, direction, amplitude, body-part involvement, and object/contact conditions.", |
| "attribution": "Jiahao Ji Affiliation: Noitom Robotics Affiliation: National University of Singapore Email: [tristan@noitomrobotics.com] Ji Ma Affiliation: Noitom Robotics Affiliation: The University of Hong Kong * Equal contribution. † Corresponding authors. Email: [lhan@noitomrobotics.com] Runhan Zhang Affiliation: Noitom Robotics Affiliation: SIGS, Tsinghua University Runyi Yu Affiliation: Noitom Robotics Affiliation: The Hong Kong University of Science and Technology Wenjia Wang Affiliation: Noitom Robotics Affiliation: The University of Hong Kong * Equal contribution. † Corresponding authors. Weiheng Chi Affiliation: Noitom Robotics Qianqian Peng Affiliation: Noitom Robotics Weichao Yan Affiliation: Noitom Robotics Yongfei Gu Affiliation: Noitom Robotics Ye Tian Affiliation: Noitom Robotics Ting Wu Affiliation: Noitom Robotics Longwei Li Affiliation: Noitom Robotics Chun Yuan Affiliation: SIGS, Tsinghua University Ruoli Dai Affiliation: Noitom Robotics Lei Han Affiliation: Noitom Robotics", |
| "images": [ |
| { |
| "src": "media/figures/hiphi_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2608.16222v2/figures/framenet.png", |
| "width": 4340, |
| "height": 1352, |
| "sha256": "5c8e1cb99f4533f73464c631395629e3a3c1735735b8b84dcfbddd96e80b013d", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| }, |
| "eventegohands": { |
| "note": "EEH-R 的真实传感器数据为事件流与灰度图。", |
| "figures": [ |
| { |
| "title": "合成事件与真实 EEH-R 数据样例", |
| "caption": "合成事件与真实 EEH-R 数据样例", |
| "source_url": "https://arxiv.org/html/2609.17189#S3.F4", |
| "source_label": "论文图 4", |
| "original_caption": "Fig. 4: Dataset samples. The top shows the synthetic dataset N-HOT3D, and the bottom shows the real-world dataset EEH-R.", |
| "attribution": "RYOSEI HARA 1 Affiliation: WATARU IKEDA 1 , , Affiliation: MASASHI HATANO 1 , , Affiliation: MARIKO ISOGAWA 1, 2 ,", |
| "images": [ |
| { |
| "src": "media/figures/eventegohands_0_0.webp", |
| "source_image_url": "https://arxiv.org/html/2609.17189v1/dataset_sample.png", |
| "width": 664, |
| "height": 540, |
| "sha256": "c893f93625cfedaf7dbc53814aaf6f17e1f3ac8e0aa3f6c44ca9c22bae9f2bf5", |
| "panel": 0 |
| } |
| ] |
| } |
| ] |
| } |
| } |
|
|
