Self-collision remains a persistent challenge in SMPL-based human pose estimation and motion generation. Under extreme articulations or stochastic motion synthesis, generated meshes frequently exhibit self-penetrations, leading to physically implausible results. We propose PoseShield, a neural collision constraint defined directly in SMPL pose space. We formulate collision correction as a constrained optimization problem and connect the learned constraint with the Eikonal equation. Enforcing Eikonal regularization ensures non-vanishing gradients near the collision boundary, improving numerical stability and robustness of the optimization process. Unlike prior methods that operate in the mesh space or rely on heuristic penalties, our approach operates directly in the low-dimensional space of human poses and is theoretically grounded. The same learned constraint extends to human motion sequences, providing a generator-agnostic post-hoc collision corrector without retraining the underlying motion model. Experiments on a newly constructed SMPL pose benchmark show that our method achieves a 95.8% success rate and outperforms state-of-the-art baselines.</p>\n","updatedAt":"2026-06-30T03:21:53.876Z","author":{"_id":"65e387095132c2edd193ae49","avatarUrl":"/avatars/39278e5b026bcbdde88c560fc54018c5.svg","fullname":"Yifan Shen","name":"SivanSX","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8610194325447083},"editors":["SivanSX"],"editorAvatarUrls":["/avatars/39278e5b026bcbdde88c560fc54018c5.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.29686","authors":[{"_id":"6a43360e763f63ca3757e95d","name":"Zhengyuan Li","hidden":false},{"_id":"6a43360e763f63ca3757e95e","name":"Zeyun Deng","hidden":false},{"_id":"6a43360e763f63ca3757e95f","name":"Yifan Shen","hidden":false},{"_id":"6a43360e763f63ca3757e960","name":"Liangyan Gui","hidden":false},{"_id":"6a43360e763f63ca3757e961","name":"Miaolan Xie","hidden":false},{"_id":"6a43360e763f63ca3757e962","name":"Joseph Campbell","hidden":false},{"_id":"6a43360e763f63ca3757e963","name":"Xifeng Gao","hidden":false},{"_id":"6a43360e763f63ca3757e964","name":"Kui Wu","hidden":false},{"_id":"6a43360e763f63ca3757e965","name":"Zherong Pan","hidden":false},{"_id":"6a43360e763f63ca3757e966","name":"Aniket Bera","hidden":false}],"publishedAt":"2026-06-29T00:00:00.000Z","submittedOnDailyAt":"2026-06-30T00:00:00.000Z","title":"PoseShield: Neural Collision Fields for Human Self-Collision Resolution","submittedOnDailyBy":{"_id":"65e387095132c2edd193ae49","avatarUrl":"/avatars/39278e5b026bcbdde88c560fc54018c5.svg","isPro":false,"fullname":"Yifan Shen","user":"SivanSX","type":"user","name":"SivanSX"},"summary":"Self-collision remains a persistent challenge in SMPL-based human pose estimation and motion generation. Under extreme articulations or stochastic motion synthesis, generated meshes frequently exhibit self-penetrations, leading to physically implausible results. We propose PoseShield, a neural collision constraint defined directly in SMPL pose space. We formulate collision correction as a constrained optimization problem and connect the learned constraint with the Eikonal equation. Enforcing Eikonal regularization ensures non-vanishing gradients near the collision boundary, improving numerical stability and robustness of the optimization process. Unlike prior methods that operate in the mesh space or rely on heuristic penalties, our approach operates directly in the low-dimensional space of human poses and is theoretically grounded. The same learned constraint extends to human motion sequences, providing a generator-agnostic post-hoc collision corrector without retraining the underlying motion model. Experiments on a newly constructed SMPL pose benchmark show that our method achieves a 95.8% success rate and outperforms state-of-the-art baselines.","upvotes":1,"discussionId":"6a43360e763f63ca3757e967","githubRepo":"https://github.com/lzhyu/PoseShield","githubRepoAddedBy":"user","ai_summary":"PoseShield addresses self-collision issues in SMPL-based human pose estimation by applying neural collision constraints in pose space through constrained optimization and Eikonal regularization.","ai_keywords":["SMPL","self-collision","pose space","constrained optimization","Eikonal equation","neural collision constraint","motion generation","pose estimation"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","githubStars":6},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"65e387095132c2edd193ae49","avatarUrl":"/avatars/39278e5b026bcbdde88c560fc54018c5.svg","isPro":false,"fullname":"Yifan Shen","user":"SivanSX","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.29686.md","query":{}}">
PoseShield: Neural Collision Fields for Human Self-Collision Resolution
Authors: ,
,
,
,
,
,
,
,
,
Abstract
PoseShield addresses self-collision issues in SMPL-based human pose estimation by applying neural collision constraints in pose space through constrained optimization and Eikonal regularization.
Self-collision remains a persistent challenge in SMPL-based human pose estimation and motion generation. Under extreme articulations or stochastic motion synthesis, generated meshes frequently exhibit self-penetrations, leading to physically implausible results. We propose PoseShield, a neural collision constraint defined directly in SMPL pose space. We formulate collision correction as a constrained optimization problem and connect the learned constraint with the Eikonal equation. Enforcing Eikonal regularization ensures non-vanishing gradients near the collision boundary, improving numerical stability and robustness of the optimization process. Unlike prior methods that operate in the mesh space or rely on heuristic penalties, our approach operates directly in the low-dimensional space of human poses and is theoretically grounded. The same learned constraint extends to human motion sequences, providing a generator-agnostic post-hoc collision corrector without retraining the underlying motion model. Experiments on a newly constructed SMPL pose benchmark show that our method achieves a 95.8% success rate and outperforms state-of-the-art baselines.
Community
Self-collision remains a persistent challenge in SMPL-based human pose estimation and motion generation. Under extreme articulations or stochastic motion synthesis, generated meshes frequently exhibit self-penetrations, leading to physically implausible results. We propose PoseShield, a neural collision constraint defined directly in SMPL pose space. We formulate collision correction as a constrained optimization problem and connect the learned constraint with the Eikonal equation. Enforcing Eikonal regularization ensures non-vanishing gradients near the collision boundary, improving numerical stability and robustness of the optimization process. Unlike prior methods that operate in the mesh space or rely on heuristic penalties, our approach operates directly in the low-dimensional space of human poses and is theoretically grounded. The same learned constraint extends to human motion sequences, providing a generator-agnostic post-hoc collision corrector without retraining the underlying motion model. Experiments on a newly constructed SMPL pose benchmark show that our method achieves a 95.8% success rate and outperforms state-of-the-art baselines.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2606.29686 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.29686 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.29686 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.