# InterDigitalInc/CompressAI

A PyTorch library and evaluation platform for end-to-end compression research

Repository: https://github.com/InterDigitalInc/CompressAI
Canonical: https://ross.abutalabs.com/products/compressai
Homepage: https://interdigitalinc.github.io/CompressAI/
Language: Python
License: BSD-3-Clause-Clear
License Family: permissive
Topics: compression, deep-learning, python, pytorch, machine-learning, deep-neural-networks, neural-network
Last push: 2026-07-04T00:31:53+00:00

## Health v2 (maintenance only)
Score: 73/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 90, release rhythm 35, longevity 100
- inputs: {"age_days": 2263, "days_push": 61, "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 1627, forks 277 (observed 2026-08-28T04:05:13.134348+00:00)

## What it is
CompressAI is a PyTorch library and evaluation platform for end-to-end learned data compression research. It provides custom layers, entropy models, pre-trained image/video compression models, and benchmarking scripts against classical codecs.

## Use cases
- train neural network models for learned image compression
- evaluate learned compression models against JPEG and other codecs
- encode and decode images with pre-trained compression models
- implement custom entropy models and latent codecs in PyTorch
- benchmark rate-distortion performance on datasets like Kodak
- port TensorFlow Compression functionality to PyTorch

## When to choose
- you are researching deep learning based image or video compression
- you need pre-trained learned compression models with a model zoo
- you want a PyTorch alternative to TensorFlow Compression
- you need standardized evaluation scripts comparing against classical codecs

## When to avoid
- you need general-purpose file or archive compression like zip or zstd
- you require production deployment on Windows or multi-GPU training at scale
- you need a non-PyTorch framework such as TensorFlow or JAX
- you need stable, long-term-supported APIs for production systems

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, compression, image-processing, video-processing, benchmarking
- domain: deep-learning, machine-learning, image-processing
- platform: python
- tags: learned-compression, pytorch, neural-codecs, image-compression, video-compression, entropy-models, model-zoo, video, research, linux, macos, gpu

## Member repositories
- InterDigitalInc/CompressAI (main) score 73

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:13.134348+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 2026-08-30T03:48:22.978411+00:00, confidence not recorded.
  - readme: https://github.com/InterDigitalInc/CompressAI (fetched 2026-08-28T04:05:13.134348+00:00, sha 6cce9c2e6fd0)
  - homepage: https://interdigitalinc.github.io/CompressAI/ (fetched 2026-08-29T11:21:00.741942+00:00, sha 2b79e65cda1e)
  - registry_pypi: https://pypi.org/pypi/compressai/json (fetched 2026-08-29T11:21:00.751768+00:00, sha be971d57b5fb)
- Data as of 2026-08-30T08:39:29.467469+00:00.