# VisDrone/VisDrone-Dataset

The dataset for drone based detection and tracking is released, including both image/video, and annotations.

Repository: https://github.com/VisDrone/VisDrone-Dataset
Canonical: https://ross.abutalabs.com/products/visdrone-dataset
License Family: other
Last push: 2023-09-24T06:42:49+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2563, "days_push": 1074, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2514, forks 243 (observed 2026-08-28T04:06:57.539936+00:00)

## What it is
VisDrone is a large-scale benchmark dataset of drone-captured images and videos with over 2.6 million annotated bounding boxes for object detection, tracking, and crowd counting tasks. It includes 288 video clips, 261,908 frames, and 10,209 static images collected across 14 cities in China under varied conditions.

## Use cases
- train object detection models on aerial drone imagery
- benchmark multi-object tracking algorithms on drone video
- evaluate single-object tracking from a first-frame annotation
- train crowd counting models on drone footage
- find a labeled dataset for small object detection research
- compare detection performance across weather and lighting conditions

## When to choose
- you need annotated aerial/drone imagery for detection or tracking research
- you want a standard benchmark to compare against published VisDrone results
- your model must handle small, dense objects like pedestrians and vehicles from above

## When to avoid
- you need ground-level or indoor imagery
- you need a permissively licensed dataset for commercial products (no license is specified)
- your task is unrelated to drone or aerial perspectives

## Facets
- artifact type: dataset
- maturity: stable
- function: computer-vision, image-processing, video-processing, machine-learning
- domain: computer-vision, machine-learning, autonomous-vehicles
- platform: cross-platform
- tags: drone, uav, object-detection, object-tracking, crowd-counting, benchmark, aerial-imagery, annotations

## Member repositories
- VisDrone/VisDrone-Dataset (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:57.539936+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 2026-08-30T02:26:13.048458+00:00, confidence not recorded.
  - readme: https://github.com/VisDrone/VisDrone-Dataset (fetched 2026-08-28T04:06:57.539936+00:00, sha f422dc125831)
- Data as of 2026-08-30T08:39:29.467469+00:00.