# yu-takagi/StableDiffusionReconstruction

Takagi and Nishimoto, CVPR 2023

Repository: https://github.com/yu-takagi/StableDiffusionReconstruction
Canonical: https://ross.abutalabs.com/products/stablediffusionreconstruction
Language: Jupyter Notebook
License: MIT
License Family: permissive
Last push: 2023-08-17T22:19:10+00:00

## Health v2 (maintenance only)
Score: 30/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 91
- inputs: {"age_days": 1282, "days_push": 1112, "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 1127, forks 70 (observed 2026-08-28T04:03:41.987485+00:00)

## What it is
Research codebase reproducing Takagi and Nishimoto's CVPR 2023 method for reconstructing images a person viewed from fMRI brain activity using Stable Diffusion. It provides end-to-end pipelines for MRI preprocessing, feature extraction, encoding-model training, and latent-diffusion-based image reconstruction, including extensions from the technical paper such as decoding text prompts from brain signals.

## Use cases
- reconstruct images from fmri brain activity
- decode what a person is seeing from brain scans
- reproduce the takagi nishimoto CVPR 2023 brain-to-image paper
- convert brain activity into images using stable diffusion
- train encoding models on the natural scenes dataset
- decode text prompts from neural activity
- visual stimulus reconstruction from fMRI data

## When to choose
- you want to reproduce or extend the CVPR 2023 high-resolution brain-decoding method
- you have NSD fMRI data and want to reconstruct viewed images or prompts
- you need a reference implementation of latent-diffusion-based visual decoding for research

## When to avoid
- you want a polished, production-ready image generation or inference tool
- you lack fMRI datasets, GPU resources, or MRI preprocessing experience
- you need actively maintained software with extensive documentation and support

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, stable-diffusion, image-processing, computer-vision
- domain: artificial-intelligence, machine-learning, deep-learning, computer-vision, image-processing, data-science
- platform: python
- tags: neuroscience, brain-decoding, fmri, mind-reading, visual-reconstruction, cvpr-2023, research-code, generative-ai, natural-scenes-dataset, jupyter, gpu, linux

## Member repositories
- yu-takagi/StableDiffusionReconstruction (main) score 30

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:41.987485+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 2026-08-30T06:39:10.397125+00:00, confidence not recorded.
  - readme: https://github.com/yu-takagi/StableDiffusionReconstruction (fetched 2026-08-28T04:03:41.987485+00:00, sha 32deb6861505)
- Data as of 2026-08-30T08:39:29.467469+00:00.