PyTorch3D — 3D Deep Learning Library
Meta FAIR's PyTorch library for deep learning with 3D meshes, point clouds, and volumetric data.
PyTorch3D — 3D Deep Learning Library
PyTorch3D is Meta FAIR's (FAIR = Fundamental AI Research) library of reusable components for deep learning with 3D data — meshes, point clouds, and volumes. It's designed to make differentiable rendering and 3D geometry operations as straightforward as conv2d, enabling robot perception, shape reconstruction, and neural radiance field (NeRF) research.
Key features
- Differentiable mesh and point cloud renderers (rasterizer + shader)
- Efficient batched 3D ops: chamfer distance, kNN, graph convolutions
- NeRF / implicit surface utilities (ray-marching, volume rendering)
- IO for common formats: OBJ, PLY, glTF, OFF
- GPU-accelerated CUDA kernels for all core ops
Quick start
conda install pytorch3d -c pytorch3d
import torch
from pytorch3d.structures import Meshes
from pytorch3d.renderer import (FoVPerspectiveCameras, RasterizationSettings,
MeshRenderer, MeshRasterizer, SoftPhongShader)
verts, faces = load_obj_verts("model.obj")
mesh = Meshes(verts=[verts], faces=[faces])
cameras = FoVPerspectiveCameras(device="cuda")
renderer = MeshRenderer(rasterizer=MeshRasterizer(cameras=cameras,
raster_settings=RasterizationSettings(image_size=256)),
shader=SoftPhongShader(cameras=cameras, device="cuda"))
image = renderer(mesh)
npx ai-supply add pytorch3d-3d-deep-learning
Curated mirror of the open-source PyTorch3D (BSD-3-Clause). Get it from the source.
Compromise signals — malicious or tampered code (leaked secrets, backdoors, a dropped executable) — reduce the score, and known dependency CVEs carry a bounded penalty (they warrant review but never QUARANTINE — update the dependency to clear). Other dangerous-by-capability traits are risk surface, expected for some capabilities. Every finding is mapped to its OWASP control below.
Findings mapped to the OWASP Top 10 for LLM Applications (2025) and the OWASP Machine Learning Security Top 10. Expand any flagged control for the exact findings — compromise reduces the score; expected/risk-surface do not, except a known CVE, which carries a small bounded penalty (high/critical → Review).
The same gate an agent runs before installing (POST /api/v1/trust/pytorch3d-3d-deep-learning/check). Click a policy:
Consume PyTorch3D — 3D Deep Learning Library programmatically. Authenticate with an API key or session — see Authorize an agent.
# Agents: CHECK BEFORE YOU INSTALL (no auth) — score, grade, level, capability manifest
curl https://ai-supply.store/api/v1/trust/pytorch3d-3d-deep-learning
# Gate against your org policy (returns { pass, violations })
curl -X POST https://ai-supply.store/api/v1/trust/pytorch3d-3d-deep-learning/check \
-H "Content-Type: application/json" \
-d '{"minGrade":"B","denyPermissions":["shell"],"denyUnknownEgress":true}'
# CLI
npx ai-supply add pytorch3d-3d-deep-learning
# REST (install → download)
curl -X POST https://ai-supply.store/api/v1/listings/pytorch3d-3d-deep-learning/install \
-H "Authorization: Bearer $AIM_KEY"
# MCP tool
install_listing({ "slug": "pytorch3d-3d-deep-learning" })OpenAPI spec →Curated mirror — latest upstream source. See the repository for tagged releases.