# mit-han-lab/efficientvit

Efficient vision foundation models for high-resolution generation and perception.

Repository: https://github.com/mit-han-lab/efficientvit
Canonical: https://ross.abutalabs.com/products/efficientvit
Language: Python
License: Apache-2.0
License Family: permissive
Topics: high-resolution, imagenet, efficientvit, segment-anything, segmentation, vision-transformer, deep-compression-autoencoder, efficient-diffusion-model
Last push: 2025-09-05T03:20:07+00:00

## Health v2 (maintenance only)
Score: 48/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 40, release rhythm 35, longevity 89
- inputs: {"age_days": 1246, "days_push": 362, "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 3354, forks 256 (observed 2026-08-28T04:07:57.879476+00:00)

## What it is
A collection of efficient vision foundation models from MIT Han Lab, including EfficientViT backbones for perception, EfficientViT-SAM for fast segmentation, and DC-AE deep compression autoencoders for high-resolution image generation. It provides pretrained models and training code in Python.

## Use cases
- run segment anything models fast on a laptop
- efficient image segmentation backbone for object detection
- compress images for high-resolution diffusion model training
- image generation on imagenet 512x512
- accelerate SAM inference
- efficient vision transformer for edge devices like jetson

## When to choose
- you need state-of-the-art efficient SAM segmentation or a fast ViT backbone
- you need a high-compression autoencoder for diffusion-based image generation
- you want pretrained efficient vision models for GPU or edge deployment

## When to avoid
- you need active maintenance or future updates - the repo is deprecated in favor of DC-Gen
- you need non-vision machine learning tasks
- you need a no-GPU CPU-only production solution

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, image-processing, computer-vision, llm-inference
- domain: computer-vision, image-processing, deep-learning, artificial-intelligence
- platform: python, cross-platform
- tags: vision-transformer, segment-anything, autoencoder, diffusion-models, efficient-inference, imagenet, high-resolution, gpu, linux

## Member repositories
- mit-han-lab/efficientvit (main) score 48

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:57.879476+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 2026-08-29T18:40:56.178876+00:00, confidence not recorded.
  - readme: https://github.com/mit-han-lab/efficientvit (fetched 2026-08-28T04:07:57.879476+00:00, sha 4b03ed2e8be7)
- Data as of 2026-08-30T08:39:29.467469+00:00.