# d-li14/involution

[CVPR 2021] Involution: Inverting the Inherence of Convolution for Visual Recognition, a brand new neural operator

Repository: https://github.com/d-li14/involution
Canonical: https://ross.abutalabs.com/products/involution
Homepage: https://arxiv.org/abs/2103.06255
Language: Python
License: MIT
License Family: permissive
Topics: involution, operator, pytorch, image-classification, object-detection, instance-segmentation, semantic-segmentation, cvpr2021, pre-trained-model
Last push: 2021-07-16T06:01:08+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2042, "days_push": 1874, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1310, forks 175 (observed 2026-08-28T04:04:19.931964+00:00)

## What it is
Official PyTorch implementation of the involution neural operator from the CVPR 2021 paper 'Involution: Inverting the Inherence of Convolution for Visual Recognition'. It provides model files and configs built on OpenMMLab toolkits (mmclassification, mmdetection, mmsegmentation) with pre-trained checkpoints for ImageNet classification, COCO detection/segmentation, and Cityscapes segmentation.

## Use cases
- replace convolution with involution in vision models
- train image classification models on ImageNet with involution backbones
- run object detection with involution-based ResNet on COCO
- semantic segmentation on Cityscapes using involution
- reproduce CVPR 2021 involution paper results
- compare involution against convolution and self-attention baselines

## When to choose
- you want a more efficient alternative to convolution or self-attention in vision backbones
- you already use OpenMMLab toolkits like mmdetection or mmsegmentation
- you need pre-trained involution checkpoints for classification, detection, or segmentation research

## When to avoid
- you need a maintained production library - the repo has not been updated since 2021
- your stack is TensorFlow, JAX, or non-OpenMMLab PyTorch frameworks
- you need plug-and-play layers without copying files into OpenMMLab codebases

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, image-processing, computer-vision
- domain: computer-vision, deep-learning, machine-learning
- platform: python
- tags: pytorch, neural-operator, involution, cvpr2021, pretrained-models, openmmlab, image-classification, object-detection, semantic-segmentation, research-code, gpu

## Member repositories
- d-li14/involution (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:19.931964+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 2026-08-30T04:49:54.480050+00:00, confidence not recorded.
  - readme: https://github.com/d-li14/involution (fetched 2026-08-28T04:04:19.931964+00:00, sha 8007d8ca5994)
  - homepage: https://arxiv.org/abs/2103.06255 (fetched 2026-08-29T12:08:02.133612+00:00, sha 7ce1c7dea437)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T12:08:02.136317+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T12:08:02.140101+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T12:08:02.141908+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T12:08:02.138048+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.