# fizyr/keras-retinanet

Keras implementation of RetinaNet object detection.

Repository: https://github.com/fizyr/keras-retinanet
Canonical: https://ross.abutalabs.com/products/keras-retinanet
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2023-03-16T14:51:24+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 3306, "days_push": 1266, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4383, forks 1914 (observed 2026-08-28T04:08:47.001830+00:00)

## What it is
A Keras/TensorFlow implementation of the RetinaNet object detection model with focal loss, supporting training and inference on custom datasets. The repository is deprecated in favor of torchvision's object detection module.

## Use cases
- train a custom object detection model in Keras
- detect objects in images with RetinaNet
- implement focal loss for dense object detection
- fine-tune RetinaNet on MS COCO
- convert a trained detection model for inference

## When to choose
- you are locked into Keras 2.4 / TensorFlow 2.3 and need RetinaNet
- you need a reference implementation of focal loss-based detection
- you want to train object detection on custom annotated datasets

## When to avoid
- starting a new project - use torchvision's detection module instead
- you need support for recent TensorFlow or Keras versions
- you need actively maintained code or bug fixes

## Facets
- artifact type: library
- maturity: abandoned
- function: machine-learning, deep-learning, computer-vision, image-processing
- domain: computer-vision, machine-learning, deep-learning
- platform: python
- tags: object-detection, retinanet, keras, tensorflow, deprecated, focal-loss, gpu

## Member repositories
- fizyr/keras-retinanet (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:47.001830+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 2026-08-29T18:21:17.114178+00:00, confidence not recorded.
  - readme: https://github.com/fizyr/keras-retinanet (fetched 2026-08-28T04:08:47.001830+00:00, sha ae3446fa1d3e)
  - registry_pypi: https://pypi.org/pypi/keras-retinanet/json (fetched 2026-08-29T09:09:25.241318+00:00, sha 19e2262fbafa)
- Data as of 2026-08-30T08:39:29.467469+00:00.