# Megvii-BaseDetection/YOLOX

YOLOX is a high-performance anchor-free YOLO, exceeding yolov3~v5 with MegEngine, ONNX, TensorRT, ncnn, and OpenVINO supported. Documentation: https://yolox.readthedocs.io/

Repository: https://github.com/Megvii-BaseDetection/YOLOX
Canonical: https://ross.abutalabs.com/products/yolox
Language: Python
License: Apache-2.0
License Family: permissive
Topics: yolox, yolov3, onnx, tensorrt, ncnn, openvino, pytorch, megengine, object-detection, yolo, deep-learning
Last push: 2025-06-08T03:52:54+00:00

## Health v2 (maintenance only)
Score: 34/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 25, release rhythm 8, longevity 100
- inputs: {"age_days": 1874, "days_push": 451, "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 10587, forks 2536 (observed 2026-08-28T04:10:42.489896+00:00)

## What it is
YOLOX is a high-performance anchor-free YOLO object detection model family implemented in PyTorch, with pretrained weights and export support for ONNX, TensorRT, ncnn, OpenVINO, and MegEngine. It bridges research and industrial use with benchmarked models from nano to extra-large sizes.

## Use cases
- train a custom object detection model on my own dataset
- detect objects in images or video in real time
- export a YOLO model to ONNX or TensorRT for deployment
- run object detection on edge devices with ncnn or OpenVINO
- compare anchor-free YOLO performance against YOLOv5
- fine-tune pretrained YOLOX weights for my use case

## When to choose
- you need state-of-the-art anchor-free YOLO detection with pretrained weights
- you want flexible deployment across ONNX, TensorRT, ncnn, and OpenVINO
- you need a PyTorch codebase that is easy to train and fine-tune

## When to avoid
- you need segmentation or pose estimation rather than bounding-box detection
- you want a maintained one-command CLI detector rather than a research codebase
- you work outside Python/PyTorch and only need a prebuilt inference binary

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, deep-learning, computer-vision, image-processing
- domain: computer-vision, deep-learning, machine-learning, artificial-intelligence
- platform: python, cross-platform
- tags: object-detection, yolo, anchor-free, pytorch, onnx, tensorrt, ncnn, openvino, megengine, model-zoo, inference, gpu, linux, docker

## Member repositories
- Megvii-BaseDetection/YOLOX (main) score 34

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:42.489896+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 2026-08-29T17:18:30.122257+00:00, confidence not recorded.
  - readme: https://github.com/Megvii-BaseDetection/YOLOX (fetched 2026-08-28T04:10:42.489896+00:00, sha 08792204b8c1)
  - registry_pypi: https://pypi.org/pypi/yolox/json (fetched 2026-08-29T08:17:34.786512+00:00, sha 55dae0903e8f)
- Data as of 2026-08-30T08:39:29.467469+00:00.