{"url":"/sota/3d-face-reconstruction-on-now-benchmark-1","task":{"name":"3D Face Reconstruction","url":"/task/3d-face-reconstruction","note":null},"dataset":{"name":"NoW Benchmark","url":"/dataset/now-benchmark"},"category":"Computer Vision","categories":["Computer Vision"],"category_note":null,"description":"**3D Face Reconstruction** is a computer vision task that involves creating a 3D model of a human face from a 2D image or a set of images. The goal of 3D face reconstruction is to reconstruct a digital 3D representation of a person's face, which can be used for various applications such as animation, virtual reality, and biometric identification.\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [3DDFA_V2](https://github.com/cleardusk/3DDFA_V2) )</span>","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["Median Reconstruction Error","Mean Reconstruction Error (mm)","Stdev Reconstruction Error (mm)"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Median Reconstruction Error":"lower","Mean Reconstruction Error (mm)":"lower","Stdev Reconstruction Error (mm)":"lower"}},"counts":{"rows":17,"rows_with_code":14,"rows_with_paper_page":17,"rows_dated":16,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"DenseLandmarks (Multi-view)","metrics":{"Mean Reconstruction Error (mm)":"1.01","Median Reconstruction Error":"0.81","Stdev Reconstruction Error (mm)":"0.84"},"uses_additional_data":false,"paper_date":"2022-04-06","paper":"/paper/3d-face-reconstruction-with-dense-landmarks","paper_url":"https://arxiv.org/abs/2204.02776v2","paper_title":"3D face reconstruction with dense landmarks","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":2,"model":"MICA","metrics":{"Mean Reconstruction Error (mm)":"1.11","Median Reconstruction Error":"0.90","Stdev Reconstruction Error (mm)":"0.92"},"uses_additional_data":false,"paper_date":"2022-04-13","paper":"/paper/towards-metrical-reconstruction-of-human","paper_url":"https://arxiv.org/abs/2204.06607v2","paper_title":"Towards Metrical Reconstruction of Human Faces","code":"https://github.com/Zielon/MICA","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"DenseLandmarks (Single-view)","metrics":{"Mean Reconstruction Error (mm)":"1.28","Median Reconstruction Error":"1.02","Stdev Reconstruction Error (mm)":"1.08"},"uses_additional_data":false,"paper_date":"2022-04-06","paper":"/paper/3d-face-reconstruction-with-dense-landmarks","paper_url":"https://arxiv.org/abs/2204.02776v2","paper_title":"3D face reconstruction with dense landmarks","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":4,"model":"FOCUS","metrics":{"Mean Reconstruction Error (mm)":"1.30","Median Reconstruction Error":"1.04","Stdev Reconstruction Error (mm)":"1.10"},"uses_additional_data":false,"paper_date":"2021-06-17","paper":"/paper/to-fit-or-not-to-fit-model-based-face","paper_url":"https://arxiv.org/abs/2106.09614v3","paper_title":"Robust Model-based Face Reconstruction through Weakly-Supervised Outlier Segmentation","code":"https://github.com/unibas-gravis/Occlusion-Robust-MoFA","n_code_links":1,"syntology":null},{"rank_in_archive_order":5,"model":"DECA","metrics":{"Mean Reconstruction Error (mm)":"1.38","Median Reconstruction Error":"1.09","Stdev Reconstruction Error (mm)":"1.18"},"uses_additional_data":false,"paper_date":"2020-12-07","paper":"/paper/learning-an-animatable-detailed-3d-face-model","paper_url":"https://arxiv.org/abs/2012.04012v2","paper_title":"Learning an Animatable Detailed 3D Face Model from In-The-Wild Images","code":"https://github.com/yfeng95/deca","n_code_links":2,"syntology":null},{"rank_in_archive_order":6,"model":"Deep3DFaceRecon PyTorch","metrics":{"Mean Reconstruction Error (mm)":"1.41","Median Reconstruction Error":"1.11","Stdev Reconstruction Error (mm)":"1.21"},"uses_additional_data":false,"paper_date":"2019-03-20","paper":"/paper/accurate-3d-face-reconstruction-with-weakly","paper_url":"https://arxiv.org/abs/1903.08527v2","paper_title":"Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set","code":"https://github.com/Microsoft/Deep3DFaceReconstruction","n_code_links":4,"syntology":{"n_ran":15,"n_unverified":21,"n_samples":36,"n_pointer_only_licence":0}},{"rank_in_archive_order":7,"model":"PIXIE","metrics":{"Mean Reconstruction Error (mm)":"1.49","Median Reconstruction Error":"1.18","Stdev Reconstruction Error (mm)":"1.25"},"uses_additional_data":false,"paper_date":"2021-05-11","paper":"/paper/collaborative-regression-of-expressive-bodies","paper_url":"https://arxiv.org/abs/2105.05301v2","paper_title":"Collaborative Regression of Expressive Bodies using Moderation","code":"https://github.com/YadiraF/PIXIE","n_code_links":1,"syntology":null},{"rank_in_archive_order":8,"model":"RingNet","metrics":{"Mean Reconstruction Error (mm)":"1.53","Median Reconstruction Error":"1.21","Stdev Reconstruction Error (mm)":"1.31"},"uses_additional_data":false,"paper_date":"2019-05-16","paper":"/paper/learning-to-regress-3d-face-shape-and","paper_url":"https://arxiv.org/abs/1905.06817v1","paper_title":"Learning to Regress 3D Face Shape and Expression from an Image without 3D Supervision","code":"https://github.com/soubhiksanyal/RingNet","n_code_links":2,"syntology":{"n_ran":5,"n_unverified":1,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":9,"model":"FLAME template","metrics":{"Mean Reconstruction Error (mm)":"1.53","Median Reconstruction Error":"1.21","Stdev Reconstruction Error (mm)":"1.31"},"uses_additional_data":false,"paper_date":"2017-11-01","paper":"/paper/learning-a-model-of-facial-shape-and","paper_url":"http://flame.is.tue.mpg.de/","paper_title":"Learning a model of facial shape and expression from 4D scans","code":"https://github.com/YadiraF/DECA","n_code_links":9,"syntology":null},{"rank_in_archive_order":10,"model":"Deng et al. 2019","metrics":{"Mean Reconstruction Error (mm)":"1.54","Median Reconstruction Error":"1.23","Stdev Reconstruction Error (mm)":"1.29"},"uses_additional_data":false,"paper_date":"2019-03-20","paper":"/paper/accurate-3d-face-reconstruction-with-weakly","paper_url":"https://arxiv.org/abs/1903.08527v2","paper_title":"Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set","code":"https://github.com/Microsoft/Deep3DFaceReconstruction","n_code_links":4,"syntology":{"n_ran":15,"n_unverified":21,"n_samples":36,"n_pointer_only_licence":0}},{"rank_in_archive_order":11,"model":"3DDFA_V2","metrics":{"Mean Reconstruction Error (mm)":"1.57","Median Reconstruction Error":"1.23","Stdev Reconstruction Error (mm)":"1.39"},"uses_additional_data":false,"paper_date":"2020-09-21","paper":"/paper/towards-fast-accurate-and-stable-3d-dense-1","paper_url":"https://arxiv.org/abs/2009.09960v2","paper_title":"Towards Fast, Accurate and Stable 3D Dense Face Alignment","code":"https://github.com/cleardusk/3DDFA","n_code_links":3,"syntology":{"n_ran":6,"n_unverified":4,"n_samples":10,"n_pointer_only_licence":3}},{"rank_in_archive_order":12,"model":"Dib et al. 2021","metrics":{"Mean Reconstruction Error (mm)":"1.57","Median Reconstruction Error":"1.26","Stdev Reconstruction Error (mm)":"1.31"},"uses_additional_data":false,"paper_date":"2021-03-29","paper":"/paper/towards-high-fidelity-monocular-face","paper_url":"https://arxiv.org/abs/2103.15432v3","paper_title":"Towards High Fidelity Monocular Face Reconstruction with Rich Reflectance using Self-supervised Learning and Ray Tracing","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":13,"model":"SynergyNet","metrics":{"Mean Reconstruction Error (mm)":"1.59","Median Reconstruction Error":"1.27","Stdev Reconstruction Error (mm)":"1.31"},"uses_additional_data":false,"paper_date":"2021-10-19","paper":"/paper/synergy-between-3dmm-and-3d-landmarks-for","paper_url":"https://arxiv.org/abs/2110.09772v3","paper_title":"Synergy between 3DMM and 3D Landmarks for Accurate 3D Facial Geometry","code":"https://github.com/tomas-gajarsky/facetorch","n_code_links":4,"syntology":null},{"rank_in_archive_order":14,"model":"MGCNet","metrics":{"Mean Reconstruction Error (mm)":"1.87","Median Reconstruction Error":"1.31","Stdev Reconstruction Error (mm)":"2.63"},"uses_additional_data":false,"paper_date":"2020-07-24","paper":"/paper/self-supervised-monocular-3d-face","paper_url":"https://arxiv.org/abs/2007.12494v1","paper_title":"Self-Supervised Monocular 3D Face Reconstruction by Occlusion-Aware Multi-view Geometry Consistency","code":"https://github.com/jiaxiangshang/MGCNet","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":16,"n_samples":16,"n_pointer_only_licence":0}},{"rank_in_archive_order":15,"model":"PRNet","metrics":{"Mean Reconstruction Error (mm)":"1.98","Median Reconstruction Error":"1.50","Stdev Reconstruction Error (mm)":"1.88"},"uses_additional_data":false,"paper_date":"2018-03-21","paper":"/paper/joint-3d-face-reconstruction-and-dense","paper_url":"http://arxiv.org/abs/1803.07835v1","paper_title":"Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network","code":"https://github.com/YadiraF/PRNet","n_code_links":4,"syntology":{"n_ran":0,"n_unverified":10,"n_samples":10,"n_pointer_only_licence":0}},{"rank_in_archive_order":16,"model":"UMDFA","metrics":{"Mean Reconstruction Error (mm)":"1.89","Median Reconstruction Error":"1.52","Stdev Reconstruction Error (mm)":"1.57"},"uses_additional_data":false,"paper_date":null,"paper":"/paper/look-ma-no-landmarks-unsupervised-model-based","paper_url":"https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/4043_ECCV_2020_paper.php","paper_title":"“Look Ma, no landmarks!” – Unsupervised, Model-based Dense Face Alignment","code":"https://github.com/kzmttr/UMDFA","n_code_links":1,"syntology":null},{"rank_in_archive_order":17,"model":"3DMM-CNN","metrics":{"Mean Reconstruction Error (mm)":"2.33","Median Reconstruction Error":"1.84","Stdev Reconstruction Error (mm)":"2.05"},"uses_additional_data":false,"paper_date":"2016-12-15","paper":"/paper/regressing-robust-and-discriminative-3d","paper_url":"http://arxiv.org/abs/1612.04904v1","paper_title":"Regressing Robust and Discriminative 3D Morphable Models with a very Deep Neural Network","code":"https://github.com/anhttran/3dmm_cnn","n_code_links":5,"syntology":null}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,885 of the 9,623 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9623,"papers_checked":6885,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2737},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-25T09:33:49+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":6,"rows_with_any_sample_ran":4,"distinct_papers_with_graph_line":5,"distinct_papers_with_any_sample_ran":3,"samples_over_distinct_papers":{"n_ran":26,"n_unverified":52,"n_samples":78,"n_pointer_only_licence":3,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":41,"n_unverified":73,"n_samples":114,"n_pointer_only_licence":3,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}