# DmitryUlyanov/deep-image-prior

Image restoration with neural networks but without learning.

Repository: https://github.com/DmitryUlyanov/deep-image-prior
Canonical: https://ross.abutalabs.com/products/deep-image-prior
Homepage: https://dmitryulyanov.github.io/deep_image_prior
Language: Jupyter Notebook
License: NOASSERTION
License Family: other
Last push: 2023-04-27T03:47:44+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": 3199, "days_push": 1224, "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 8088, forks 1440 (observed 2026-08-28T04:10:12.633494+00:00)

## What it is
Jupyter Notebook reference implementations of the CVPR 2018 'Deep Image Prior' paper, which restores images using randomly-initialized neural networks as a structured prior without any training. It includes notebooks reproducing denoising, super-resolution, and inpainting results, runnable locally, via Docker, or on Google Colab.

## Use cases
- remove noise from images without training a model
- inpaint missing regions of a photo
- upscale images with super-resolution
- reproduce deep image prior paper figures
- restore images without a training dataset
- experiment with neural network priors for inverse problems

## When to choose
- you want training-free image restoration using a random network prior
- you want to reproduce or build on the Deep Image Prior paper
- you have a GPU and prefer notebook-based experimentation

## When to avoid
- you need a production-ready image restoration library or API
- you want fast inference - the method is slow per-image optimization
- you need guaranteed convergence on all GPUs, as some Tesla GPUs have known issues

## Facets
- artifact type: learning-resource
- maturity: maintenance
- function: image-processing, machine-learning, deep-learning
- domain: computer-vision, image-processing, deep-learning, artificial-intelligence
- platform: python, cross-platform
- tags: image-restoration, denoising, inpainting, super-resolution, jupyter-notebooks, research-code, pytorch, cvpr-2018, gpu, docker

## Member repositories
- DmitryUlyanov/deep-image-prior (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:12.633494+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 2026-08-29T17:30:47.808699+00:00, confidence not recorded.
  - readme: https://github.com/DmitryUlyanov/deep-image-prior (fetched 2026-08-28T04:10:12.633494+00:00, sha 39f782973507)
  - homepage: https://dmitryulyanov.github.io/deep_image_prior (fetched 2026-08-29T08:28:47.750808+00:00, sha 961f9293c425)
- Data as of 2026-08-30T08:39:29.467469+00:00.