# CASIA-IVA-Lab/DANet

Dual Attention Network for Scene Segmentation (CVPR2019)

Repository: https://github.com/CASIA-IVA-Lab/DANet
Canonical: https://ross.abutalabs.com/products/danet
Language: Python
License: MIT
License Family: permissive
Last push: 2024-12-23T11:42:23+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": 2913, "days_push": 618, "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 2463, forks 480 (observed 2026-08-28T04:06:53.544405+00:00)

## What it is
DANet is the official PyTorch implementation of 'Dual Attention Network for Scene Segmentation' (CVPR 2019), which uses position and channel attention to integrate local features with global dependencies. It provides training and evaluation code with pretrained models for semantic segmentation benchmarks like Cityscapes, PASCAL Context, and COCO Stuff-10k.

## Use cases
- run semantic segmentation on cityscapes images
- reproduce CVPR 2019 scene segmentation results
- train an attention-based segmentation model on my own dataset
- evaluate a pretrained DANet-101 model on cityscapes val set
- compare self-attention segmentation models
- segment street scenes for autonomous driving research

## When to choose
- you need a proven attention-based semantic segmentation baseline with pretrained weights
- you want to reproduce or extend the DANet/DRANet paper results on Cityscapes
- you are researching self-attention mechanisms in dense prediction

## When to avoid
- you need a maintained general-purpose segmentation toolkit (use MMSegmentation instead)
- you need real-time or lightweight segmentation on edge devices
- you work outside PyTorch or need modern architecture support out of the box

## 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: semantic-segmentation, scene-segmentation, self-attention, pytorch, cvpr2019, cityscapes, resnet, gpu, linux

## Member repositories
- CASIA-IVA-Lab/DANet (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:53.544405+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 2026-08-30T02:29:16.504224+00:00, confidence not recorded.
  - readme: https://github.com/CASIA-IVA-Lab/DANet (fetched 2026-08-28T04:06:53.544405+00:00, sha b4621740613d)
- Data as of 2026-08-30T08:39:29.467469+00:00.