# guochengqian/PointNeXt

[NeurIPS'22] PointNeXt: Revisiting PointNet++ with Improved Training and Scaling Strategies

Repository: https://github.com/guochengqian/PointNeXt
Canonical: https://ross.abutalabs.com/products/pointnext
Homepage: https://guochengqian.github.io/PointNeXt/
Language: Python
License: MIT
License Family: permissive
Topics: 3d-vision, pointcloud, pointnet2, pytorch
Last push: 2026-07-06T06:52:32+00:00

## Health v2 (maintenance only)
Score: 93/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 91, release rhythm 92, longevity 100
- inputs: {"age_days": 1565, "days_push": 58, "days_rel": 58, "gap_med": 0.0, "n_releases_24m": 3}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1067, forks 163 (observed 2026-08-28T04:03:27.231510+00:00)

## What it is
PointNeXt is the official PyTorch implementation of the NeurIPS'22 paper that improves PointNet++ via better training and model scaling strategies for point cloud understanding. It includes the OpenPoints library, a flexible codebase supporting many point-based networks with reproducible, fair benchmarking.

## Use cases
- train point cloud classification models
- benchmark point-based networks like PointNet++ and DGCNN fairly
- run semantic segmentation on 3D point clouds
- reproduce PointNeXt paper results with pretrained models
- build custom point cloud networks from config files
- apply improved training strategies to boost PointNet++ accuracy

## When to choose
- you need state-of-the-art point cloud classification or segmentation in PyTorch
- you want a reproducible benchmarking framework for point-based methods
- you want pretrained models and training logs for 3D vision research

## When to avoid
- you work with images or text rather than 3D point clouds
- you need a production-ready inference service rather than a research codebase
- you don't have GPU resources for training deep learning models

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, deep-learning
- domain: computer-vision, deep-learning, machine-learning
- platform: python
- tags: point-cloud, pointnet, pytorch, 3d-vision, research-code, benchmarking, neurips, gpu, linux

## Member repositories
- guochengqian/PointNeXt (main) score 93

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:27.231510+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:55:10.288641+00:00, confidence not recorded.
  - readme: https://github.com/guochengqian/PointNeXt (fetched 2026-08-28T04:03:27.231510+00:00, sha f0636ded5782)
  - homepage: https://guochengqian.github.io/PointNeXt/ (fetched 2026-08-29T12:57:15.888279+00:00, sha 50b2ed1a9abf)
- Data as of 2026-08-30T08:39:29.467469+00:00.