# zhanghang1989/ResNeSt

ResNeSt: Split-Attention Networks

Repository: https://github.com/zhanghang1989/ResNeSt
Canonical: https://ross.abutalabs.com/products/resnest
Homepage: https://arxiv.org/abs/2004.08955
Language: Python
License: Apache-2.0
License Family: permissive
Topics: deep-learning, resnet, resnest, pytorch, detectron-models, split-attention-networks
Last push: 2022-12-09T10:57:09+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 2362, "days_push": 1363, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3261, forks 495 (observed 2026-08-28T04:07:52.662792+00:00)

## What it is
ResNeSt is a PyTorch implementation of the Split-Attention Network, a ResNet variant that applies channel-wise attention across network branches for improved visual recognition. It ships pretrained ImageNet models and serves as a drop-in backbone boosting downstream tasks like Mask R-CNN, Cascade R-CNN, and DeepLabV3.

## Use cases
- image classification backbone with pretrained weights
- object detection on COCO with a stronger ResNet backbone
- semantic segmentation on ADE20K or Cityscapes
- instance and panoptic segmentation with Mask R-CNN
- transfer learning for custom vision tasks
- compare ResNet variants like EfficientNet vs ResNeSt

## When to choose
- you need a strong CNN backbone for detection or segmentation in PyTorch
- you want pretrained ImageNet weights with proven COCO/LVIS challenge results
- you want a drop-in ResNet replacement with better accuracy-latency tradeoff

## When to avoid
- you need a transformer-based vision model like ViT or Swin
- you need a framework-agnostic solution outside PyTorch
- you need actively maintained code with recent updates

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, image-processing, computer-vision
- domain: deep-learning, computer-vision, image-processing
- platform: python
- tags: resnet-variant, split-attention, pytorch, pretrained-models, backbone-network, image-classification, object-detection, semantic-segmentation, gpu

## Member repositories
- zhanghang1989/ResNeSt (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:52.662792+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 2026-08-30T07:23:13.486103+00:00, confidence not recorded.
  - readme: https://github.com/zhanghang1989/ResNeSt (fetched 2026-08-28T04:07:52.662792+00:00, sha 0a351c80b344)
  - homepage: https://arxiv.org/abs/2004.08955 (fetched 2026-08-29T09:36:50.093909+00:00, sha c51137a9a245)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T09:36:50.096501+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T09:36:50.100143+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T09:36:50.102207+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T09:36:50.098332+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.