# qubvel-org/segmentation_models.pytorch

Semantic segmentation models with 500+ pretrained convolutional and transformer-based backbones.

Repository: https://github.com/qubvel-org/segmentation_models.pytorch
Canonical: https://ross.abutalabs.com/products/segmentation_modelspytorch
Homepage: https://smp.readthedocs.io/
Language: Python
License: MIT
License Family: permissive
Topics: segmentation, image-processing, pspnet, unet, unet-pytorch, pytorch, fpn, models, imagenet, semantic-segmentation, image-segmentation, segmentation-models, unetplusplus, deeplabv3, deeplab-v3-plus, pretrained-weights, computer-vision, segformer, transformers, dpt
Last push: 2026-08-24T06:01:10+00:00

## Health v2 (maintenance only)
Score: 70/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 16, longevity 100
- inputs: {"age_days": 2742, "days_push": 9, "days_rel": 503, "gap_med": 98, "n_releases_24m": 2}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 11706, forks 1841 (observed 2026-08-28T04:10:49.535593+00:00)

## What it is
A PyTorch library providing neural networks for image semantic segmentation with a simple high-level API. It includes 12 encoder-decoder architectures (Unet, Unet++, DeepLabV3+, Segformer, DPT, etc.) and 800+ pretrained convolutional and transformer-based encoders, plus common losses and metrics.

## Use cases
- train a semantic segmentation model in pytorch
- segment images with a pretrained unet
- fine-tune a segmentation model with pretrained encoders
- use segformer or deeplabv3+ for image segmentation
- export a segmentation model to onnx
- use dice or jaccard loss for segmentation training

## When to choose
- you need ready-made segmentation architectures with pretrained backbones in PyTorch
- you want a minimal two-line API to build segmentation networks
- you need timm encoder support and ONNX/TorchScript export

## When to avoid
- you need object detection or classification rather than pixel-level segmentation
- you work outside PyTorch (e.g., TensorFlow or JAX)
- you need fully custom architectures not covered by the 12 provided models

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, deep-learning, image-processing, computer-vision
- domain: computer-vision, deep-learning, machine-learning, image-processing
- platform: python, cross-platform
- tags: semantic-segmentation, pytorch, unet, pretrained-encoders, timm, onnx-export, segmentation-models, encoder-decoder, gpu

## Member repositories
- qubvel-org/segmentation_models.pytorch (main) score 70

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:49.535593+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 2026-08-29T17:15:23.402852+00:00, confidence not recorded.
  - readme: https://github.com/qubvel-org/segmentation_models.pytorch (fetched 2026-08-28T04:10:49.535593+00:00, sha 506e2db25919)
  - homepage: https://smp.readthedocs.io/ (fetched 2026-08-29T08:13:37.994765+00:00, sha 95da24b1f460)
  - registry_pypi: https://pypi.org/pypi/segmentation_models.pytorch/json (fetched 2026-08-29T08:13:38.004417+00:00, sha 2e63681681a1)
- Data as of 2026-08-30T08:39:29.467469+00:00.