# knazeri/edge-connect

EdgeConnect: Structure Guided Image Inpainting using Edge Prediction, ICCV  2019 https://arxiv.org/abs/1901.00212

Repository: https://github.com/knazeri/edge-connect
Canonical: https://ross.abutalabs.com/products/edge-connect
Homepage: http://openaccess.thecvf.com/content_ICCVW_2019/html/AIM/Nazeri_EdgeConnect_Structure_Guided_Image_Inpainting_using_Edge_Prediction_ICCVW_2019_paper.html
Language: Python
License: NOASSERTION
License Family: other
Topics: edgeconnect, image-inpainting, generative-adversarial-network, gans, canny-edge-detection, deep-learning, pytorch
Last push: 2024-02-03T16:03:43+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": 2818, "days_push": 942, "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 2620, forks 528 (observed 2026-08-28T04:07:04.986893+00:00)

## What it is
EdgeConnect is a PyTorch implementation of a two-stage generative adversarial model for image inpainting, published at ICCV 2019. It first hallucinates edges of missing image regions with an edge generator network, then fills the regions using those edges as structural priors.

## Use cases
- fill in missing or masked regions of photos
- remove unwanted objects from images
- repair old or damaged photographs
- inpaint irregular holes in images with realistic structure
- reproduce research results from the EdgeConnect ICCV 2019 paper
- train a custom inpainting model on Places2 or CelebA

## When to choose
- you need research-grade image inpainting with strong structural detail in filled regions
- you want a PyTorch codebase with pre-trained models for Places2 and CelebA
- you have an NVIDIA GPU and want to train or fine-tune an inpainting GAN
- you need to handle both regular and irregular missing-region masks

## When to avoid
- you need a production-ready, actively maintained tool with a polished API or GUI
- you want inpainting on CPU or without CUDA-capable hardware
- you need a license-clear solution for commercial use (license is non-standard)
- you need text-guided or diffusion-based inpainting rather than mask-based inpainting

## Facets
- artifact type: library
- maturity: maintenance
- function: image-processing, computer-vision, deep-learning, machine-learning
- domain: computer-vision, image-processing, deep-learning, machine-learning
- platform: python, cross-platform
- tags: image-inpainting, gan, pytorch, edge-prediction, iccv-2019, research-code, canny-edge-detection, gpu, linux

## Member repositories
- knazeri/edge-connect (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:04.986893+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 2026-08-30T02:20:11.625841+00:00, confidence not recorded.
  - readme: https://github.com/knazeri/edge-connect (fetched 2026-08-28T04:07:04.986893+00:00, sha 802703b52fea)
  - homepage: http://openaccess.thecvf.com/content_ICCVW_2019/html/AIM/Nazeri_EdgeConnect_Structure_Guided_Image_Inpainting_using_Edge_Prediction_ICCVW_2019_paper.html (fetched 2026-08-29T10:03:11.038652+00:00, sha 354370030bfe)
- Data as of 2026-08-30T08:39:29.467469+00:00.