# roboflow/inference

Turn any computer or edge device into a command center for your computer vision projects.

Repository: https://github.com/roboflow/inference
Canonical: https://ross.abutalabs.com/products/roboflow-inference
Homepage: https://inference.roboflow.com
Language: Python
License: NOASSERTION
License Family: other
Topics: computer-vision, inference-api, inference-server, vit, yolov5, yolov8, jetson, tensorrt, classification, instance-segmentation, object-detection, onnx, deployment, docker, inference, machine-learning, python, yolo11, agents, yolov12
Last push: 2026-08-26T20:52:12+00:00

## Health v2 (maintenance only)
Score: 91/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 86, longevity 80
- inputs: {"age_days": 1129, "days_push": 7, "days_rel": 12, "gap_med": 4.0, "n_releases_24m": 151}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2427, forks 310 (observed 2026-08-28T04:06:50.884749+00:00)

## What it is
Roboflow Inference is a Python library and self-hostable inference server for deploying computer vision models on any computer or edge device. It supports fine-tuned models, foundation models like Florence-2, CLIP, and SAM2, and a Workflows engine for chaining models, logic, and integrations into production pipelines.

## Use cases
- deploy a fine-tuned object detection model on an edge device
- run YOLOv8 inference on live RTSP camera streams
- self-host a computer vision inference API server
- build multi-model pipelines with CLIP and SAM2
- count and track objects in video with notifications
- blur faces or read license plates from camera feeds
- run OCR and barcode detection alongside ML models

## When to choose
- you need to deploy vision models on-premise or at the edge with optional offline operation
- you want to chain multiple models and CV techniques into production pipelines
- you need GPU-accelerated inference served over an HTTP API
- you want pre-built blocks for tracking, counting, and video stream management

## When to avoid
- you only need training or fine-tuning of models rather than deployment
- you need a lightweight single-model library without server infrastructure
- your project requires a permissive license without restrictions (license is custom)
- you want a fully managed cloud service without self-hosting

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, computer-vision, llm-inference, video-processing, ocr, sdk, self-hosted, api-framework
- domain: computer-vision, machine-learning, deep-learning, image-processing, developer-tools
- platform: python, windows, self-hosted, cloud
- tags: inference-server, object-detection, yolo, edge-deployment, foundation-models, workflows, rtsp, sam2, clip, model-serving, video, docker, linux, macos, gpu

## Member repositories
- roboflow/inference (main) score 91

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:50.884749+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 2026-08-30T02:31:36.330070+00:00, confidence not recorded.
  - readme: https://github.com/roboflow/inference (fetched 2026-08-28T04:06:50.884749+00:00, sha 1b0ed7a89ea6)
  - homepage: https://inference.roboflow.com (fetched 2026-08-29T10:12:56.136324+00:00, sha f37ec2985107)
  - site_page: https://docs.roboflow.com/workflows (fetched 2026-08-29T10:12:56.145531+00:00, sha 76029299e094)
  - registry_pypi: https://pypi.org/pypi/inference/json (fetched 2026-08-29T10:12:56.147216+00:00, sha ec30d4ecffba)
- Data as of 2026-08-30T08:39:29.467469+00:00.