# apple/ml-cvnets

CVNets: A library for training computer vision networks

Repository: https://github.com/apple/ml-cvnets
Canonical: https://ross.abutalabs.com/products/ml-cvnets
Homepage: https://apple.github.io/ml-cvnets
Language: Python
License: NOASSERTION
License Family: other
Topics: ade20k, classification, computer-vision, deep-learning, detection, imagenet, machine-learning, mscoco, pascal-voc, pytorch, segmentation
Last push: 2023-10-30T17:05:10+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 1777, "days_push": 1038, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1983, forks 258 (observed 2026-08-28T04:06:01.928103+00:00)

## What it is
CVNets is Apple's open-source PyTorch library for training computer vision networks, covering classification, detection, segmentation, video classification, and foundation models like CLIP. It includes implementations of many state-of-the-art mobile and non-mobile architectures (MobileNet, EfficientNet, ViT, MobileViT, Swin) plus data loading, augmentation, and sampling tools.

## Use cases
- train an image classification model on ImageNet
- train object detection models on MS-COCO
- train semantic segmentation models like DeepLabv3
- train CLIP-style foundation models
- train mobile-friendly vision transformers like MobileViT
- convert trained PyTorch vision models to CoreML
- train video classification models on Kinetics

## When to choose
- you need to train or reproduce vision models from Apple's research papers (MobileViT, RangeAugment, etc.)
- you want a single PyTorch toolkit spanning classification, detection, and segmentation
- you target mobile deployment and need efficient vision architectures with CoreML export

## When to avoid
- you only need pretrained models for inference rather than training
- you prefer higher-level frameworks like torchvision, timm, or MMDetection with larger communities
- you need active development or timely updates, as the project has not seen recent releases

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, image-processing, computer-vision
- domain: computer-vision, deep-learning, machine-learning, image-processing
- platform: python, cross-platform
- tags: pytorch, image-classification, object-detection, semantic-segmentation, clip, model-training, mobile-vision-models, coreml, imagenet, mscoco, gpu

## Member repositories
- apple/ml-cvnets (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:01.928103+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 2026-08-30T03:03:25.869790+00:00, confidence not recorded.
  - readme: https://github.com/apple/ml-cvnets (fetched 2026-08-28T04:06:01.928103+00:00, sha f40cff3b84c0)
  - homepage: https://apple.github.io/ml-cvnets (fetched 2026-08-29T10:43:08.009324+00:00, sha 710a73cad634)
- Data as of 2026-08-30T08:39:29.467469+00:00.