Learning representations of CAD models is a largely open problem. While 3D representation learning has flourished around point clouds and meshes, the native format of CAD - boundary representations BReps, which encodes exact parametric surfaces, curves, and their topology, has received little attention as a representation learning substrate. We introduce BRepCLIP, the first framework to align BRep geometry with language and image embeddings through contrastive pretraining. We model each CAD object as a sequence of face and edge tokens with separate discrete vocabularies for surface and curve geometry, augmented with spatial and semantic descriptors that capture surface types (e.g., cylindrical, torus, NURBS) and curve primitives (e.g., line, arc, B-spline). A transformer encoder aggregates these tokens into a global BRep embedding, aligned with CLIP's text and image encoders via a joint contrastive objective. BRepCLIP generates more discriminative and semantically grounded embeddings than existing point-based alternatives, improving Top-1 retrieval over OpenShape by 40.4%, 22.0%, and 23.9% on ABC, CADParser, and Automate, respectively, and improving zero-shot classification on FabWave by 15% in Top-1 score. We further demonstrate its utility as a CAD-aware similarity metric for evaluating text and image-conditioned CAD generation, establishing the importance of structure-aware pretraining for multimodal CAD understanding.</p>\n","updatedAt":"2026-06-05T19:59:39.817Z","author":{"_id":"65e07fb40d364b871d406648","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/65e07fb40d364b871d406648/ye6D56YpOyL8bHoTVmFcq.png","fullname":"Mohammad Sadil Khan","name":"SadilKhan","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":22,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8993567824363708},"editors":["SadilKhan"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/65e07fb40d364b871d406648/ye6D56YpOyL8bHoTVmFcq.png"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.05515","authors":[{"_id":"6a232a5de4c258a0294917c0","name":"Muhammad Usama","hidden":false},{"_id":"6a232a5de4c258a0294917c1","name":"Didier Stricker","hidden":false},{"_id":"6a232a5de4c258a0294917c2","name":"Mohammad Sadil Khan","hidden":false},{"_id":"6a232a5de4c258a0294917c3","name":"Muhammad Zeshan Afzal","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/65e07fb40d364b871d406648/7Vpm_oRgl4FBWdpGD6gwP.png"],"publishedAt":"2026-06-03T00:00:00.000Z","submittedOnDailyAt":"2026-06-05T00:00:00.000Z","title":"BRepCLIP: Contrastive Multimodal Pretraining on BRep Primitives for CAD Understanding","submittedOnDailyBy":{"_id":"65e07fb40d364b871d406648","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/65e07fb40d364b871d406648/ye6D56YpOyL8bHoTVmFcq.png","isPro":false,"fullname":"Mohammad Sadil Khan","user":"SadilKhan","type":"user","name":"SadilKhan"},"summary":"Learning representations of CAD models is a largely open problem. While 3D representation learning has flourished around point clouds and meshes, the native format of CAD - boundary representations BReps, which encodes exact parametric surfaces, curves, and their topology, has received little attention as a representation learning substrate. We introduce BRepCLIP, the first framework to align BRep geometry with language and image embeddings through contrastive pretraining. We model each CAD object as a sequence of face and edge tokens with separate discrete vocabularies for surface and curve geometry, augmented with spatial and semantic descriptors that capture surface types (e.g., cylindrical, torus, NURBS) and curve primitives (e.g., line, arc, B-spline). A transformer encoder aggregates these tokens into a global BRep embedding, aligned with CLIP's text and image encoders via a joint contrastive objective. BRepCLIP generates more discriminative and semantically grounded embeddings than existing point-based alternatives, improving Top-1 retrieval over OpenShape by 40.4%, 22.0%, and 23.9% on ABC, CADParser, and Automate, respectively, and improving zero-shot classification on FabWave by 15% in Top-1 score. We further demonstrate its utility as a CAD-aware similarity metric for evaluating text and image-conditioned CAD generation, establishing the importance of structure-aware pretraining for multimodal CAD understanding. Project page is available at https://muhammadusama100.github.io/BrepClip2026/","upvotes":0,"discussionId":"6a232a5ee4c258a0294917c4","projectPage":"https://muhammadusama100.github.io/BrepClip2026/","ai_summary":"BRepCLIP enables multimodal representation learning for CAD models by aligning boundary representation geometry with language and image embeddings through contrastive pretraining, achieving superior retrieval and classification performance compared to point-based methods.","ai_keywords":["BReps","boundary representations","CAD models","CLIP","contrastive pretraining","transformer encoder","face tokens","edge tokens","discrete vocabularies","spatial descriptors","semantic descriptors","global BRep embedding","multimodal CAD understanding","Top-1 retrieval","zero-shot classification"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","organization":{"_id":"65a7b8f2de63b063e319ea71","name":"DFKI","fullname":"German Research Center for Artificial Intelligence (DFKI)","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/5efda656ff69163f6f59e5d2/2m14xWaVEY0YT0iYhYUAQ.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[],"acceptLanguages":["en"],"organization":{"_id":"65a7b8f2de63b063e319ea71","name":"DFKI","fullname":"German Research Center for Artificial Intelligence (DFKI)","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/5efda656ff69163f6f59e5d2/2m14xWaVEY0YT0iYhYUAQ.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.05515.md"}">
BRepCLIP: Contrastive Multimodal Pretraining on BRep Primitives for CAD Understanding
Abstract
BRepCLIP enables multimodal representation learning for CAD models by aligning boundary representation geometry with language and image embeddings through contrastive pretraining, achieving superior retrieval and classification performance compared to point-based methods.
Learning representations of CAD models is a largely open problem. While 3D representation learning has flourished around point clouds and meshes, the native format of CAD - boundary representations BReps, which encodes exact parametric surfaces, curves, and their topology, has received little attention as a representation learning substrate. We introduce BRepCLIP, the first framework to align BRep geometry with language and image embeddings through contrastive pretraining. We model each CAD object as a sequence of face and edge tokens with separate discrete vocabularies for surface and curve geometry, augmented with spatial and semantic descriptors that capture surface types (e.g., cylindrical, torus, NURBS) and curve primitives (e.g., line, arc, B-spline). A transformer encoder aggregates these tokens into a global BRep embedding, aligned with CLIP's text and image encoders via a joint contrastive objective. BRepCLIP generates more discriminative and semantically grounded embeddings than existing point-based alternatives, improving Top-1 retrieval over OpenShape by 40.4%, 22.0%, and 23.9% on ABC, CADParser, and Automate, respectively, and improving zero-shot classification on FabWave by 15% in Top-1 score. We further demonstrate its utility as a CAD-aware similarity metric for evaluating text and image-conditioned CAD generation, establishing the importance of structure-aware pretraining for multimodal CAD understanding. Project page is available at https://muhammadusama100.github.io/BrepClip2026/
Community
Learning representations of CAD models is a largely open problem. While 3D representation learning has flourished around point clouds and meshes, the native format of CAD - boundary representations BReps, which encodes exact parametric surfaces, curves, and their topology, has received little attention as a representation learning substrate. We introduce BRepCLIP, the first framework to align BRep geometry with language and image embeddings through contrastive pretraining. We model each CAD object as a sequence of face and edge tokens with separate discrete vocabularies for surface and curve geometry, augmented with spatial and semantic descriptors that capture surface types (e.g., cylindrical, torus, NURBS) and curve primitives (e.g., line, arc, B-spline). A transformer encoder aggregates these tokens into a global BRep embedding, aligned with CLIP's text and image encoders via a joint contrastive objective. BRepCLIP generates more discriminative and semantically grounded embeddings than existing point-based alternatives, improving Top-1 retrieval over OpenShape by 40.4%, 22.0%, and 23.9% on ABC, CADParser, and Automate, respectively, and improving zero-shot classification on FabWave by 15% in Top-1 score. We further demonstrate its utility as a CAD-aware similarity metric for evaluating text and image-conditioned CAD generation, establishing the importance of structure-aware pretraining for multimodal CAD understanding.
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/2606.05515 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.05515 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.05515 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.