Data checked 2026-09-13. Data checked within the review cycle.
Review cycles are documented on the Methodology page. Methodology
VisDrone Dataset
Dataset Tier A perception_ai
The dataset for drone based detection and tracking is released, including both image/video, and annotations.
Engineering Snapshot
Use cases & tasks 6
- Best suited for
-
- Training and evaluating aerial object detectors
- Multi-object tracking from drone video
- Domain adaptation from ground to aerial imagery
- Primary tasks
-
- Drone-based object detection benchmarking
- Aerial multi-object tracking research
- Small object detection in high-resolution aerial images
Stack & ecosystem
- Resource type
- Dataset
- Ecosystem
- perception_ai
License & compliance
- License
- Unknown
Lifecycle & freshness Show
- Maintenance
- Active
- Latest version
- Not recorded
- Last activity
- 2026-09-12
- Last checked
- 2026-09-13
- Verification
- Official confirmed
What It Solves
Drone-based perception requires large-scale annotated aerial imagery for object detection and tracking under varying altitudes, viewpoints, and lighting conditions.
Primary use cases
- Drone-based object detection benchmarking
- Aerial multi-object tracking research
- Small object detection in high-resolution aerial images
Secondary use cases
- Transfer learning for UAV perception stacks
- Synthetic-to-real domain adaptation experiments
- Crowd and vehicle counting from aerial views
When to Use
Consider when
- License terms are unspecified; verify before commercial use
- Annotation format may require conversion for some frameworks
- Class distribution is skewed toward pedestrian and vehicle categories
Verify before adopting
- Exact license terms from the VisDrone website or challenge page
- Dataset splits (train/val/test) and annotation quality
- Compatibility with target detection/tracking pipeline input formats
Where It Fits
Stack layer data
Start Here
Adoption Checklist
- Needs verification Exact license terms from the VisDrone website or challenge page
- Needs verification Dataset splits (train/val/test) and annotation quality
- Needs verification Compatibility with target detection/tracking pipeline input formats
Each check stays "needs verification" until an official source confirms it; unconfirmed items are never marked verified.
Known Limitations & Unknowns
Known limitations
- License not declared in repository metadata
- No standardized datasheet or dataset card provided
- Class imbalance toward person and vehicle categories
Not publicly verified
- Exact image/video counts and resolution distribution
- Annotation format specification (COCO, MOT, custom)
- Official train/val/test split definitions
- Whether the dataset includes sequences for tracking or only frames for detection
Alternatives & Related Tools
Related tools
- ultralytics — Used for
How is it used?
Start from the recorded entry points below, then validate against the technical checklist.
- Official site https://github.com/VisDrone/VisDrone-Dataset
- Repository https://github.com/VisDrone/VisDrone-Dataset
- Documentation https://github.com/VisDrone/VisDrone-Dataset
Technical checklist
- NEEDS REVIEW License identified No license field in any source response
- OK Maintenance signal Active
- OK Verification status Official confirmed
- OK Source evidence attached 1 source record(s)
- NEEDS REVIEW Latest version recorded Not recorded
Official Links
Metadata & Governance
| License | License: Unknown — No license field in any source response. |
|---|---|
| Commercial model | Unknown |
| Maintenance status | Active |
| Verification status | Official confirmed — Confirmed via the official repository API responses in SourceRefs below. |
| Latest version | Not recorded |
| Latest release | Not recorded |
| Last activity | 2026-09-12 |
| Last checked | 2026-09-13 |
| First seen | Not recorded |
Related Resources & Dependencies
- ultralytics — used for (verified)
Recent Activity
New resource: VisDrone/VisDrone-Dataset
New resource added to the OpenFly index.
Related Knowledge
Guides
- Reviewing Aerial AI and Dataset Sources — A provenance-first review flow for aerial datasets and model repositories.
- Building an Aerial Dataset Pipeline — A step-by-step path for assembling an aerial detection dataset pipeline from an indexed dataset and model framework.
- Verifying Whether an AI Model Can Deploy to an Edge Platform — A checklist for checking model export support on a target edge platform without inventing performance claims.
- VisDrone Dataset Engineering & Benchmark Guide — An engineering guide for loading, converting, and benchmarking the VisDrone aerial object detection dataset with YOLO models.
Collections
- Aerial Mapping and Survey Stack — An open-source stack for aerial mapping work from imagery and datasets through photogrammetry processing.
- Aerial Perception Models and Datasets — Source-backed aerial dataset and model repositories for detection and tracking research and prototyping.
- Edge AI Perception Starter — A starting stack for prototyping aerial perception models before edge deployment work.
- Visual Perception Models and Data — Indexed model and dataset repositories for aerial visual perception research and prototyping.