Skip to content
NAMEFRAMECommercial PreviewApply for Pilot
Apply for Pilot
Barnyard

The barnyard capture.

A different map and a very different camera policy from the airbase: near-nadir, low altitude, seven classes, and a run the validator graded 78.3 out of 100. It is here because a C-grade run is more useful to look at than an A-grade one — the reasons it lost points are the reasons synthetic datasets usually disappoint.

Map
Map_BarnEnvironment
Frames in the run
300
Labelled instances
14,962
Resolution
1280×960
Seed
20260814
Validator grade
C · 78.3/100

What it is for

Small-object detection from low-altitude drone-like viewpoints, where objects are seen from directly above and occupy few pixels.

How the labels were made

The same engine-derived path as every NameFrame run: labels come from the per-instance ID buffer and the camera that rendered the frame, and the run carries the report and validation output it was graded on.

Engine
Unreal Engine 5.8
Camera
Near-nadir, 18 to 45 metres altitude, 1280×960
Classes
person, redbarrel, traffic_cone, box, haybayle, tractor, bucket
Annotation formats
YOLO, COCO, per-instance ID buffers, per-frame metadata JSON
Packaged
12 of 300 frames, across 1 archive
Version
1.0

What the frames look like

Frames from across the run, downscaled for the web. Nothing here was selected for looking good; they are the report previews the capture wrote.

Near-nadir frame plugin_000000 of the Unreal Engine farmyard capture, with 56 labelled instances outlined by NameFrameNear-nadir frame plugin_000023 of the Unreal Engine farmyard capture, with 61 labelled instances outlined by NameFrameNear-nadir frame plugin_000046 of the Unreal Engine farmyard capture, with 68 labelled instances outlined by NameFrameNear-nadir frame plugin_000069 of the Unreal Engine farmyard capture, with 40 labelled instances outlined by NameFrameNear-nadir frame plugin_000092 of the Unreal Engine farmyard capture, with 42 labelled instances outlined by NameFrameNear-nadir frame plugin_000115 of the Unreal Engine farmyard capture, with 28 labelled instances outlined by NameFrameNear-nadir frame plugin_000138 of the Unreal Engine farmyard capture, with 37 labelled instances outlined by NameFrameNear-nadir frame plugin_000161 of the Unreal Engine farmyard capture, with 54 labelled instances outlined by NameFrameNear-nadir frame plugin_000184 of the Unreal Engine farmyard capture, with 91 labelled instances outlined by NameFrameNear-nadir frame plugin_000207 of the Unreal Engine farmyard capture, with 41 labelled instances outlined by NameFrameNear-nadir frame plugin_000230 of the Unreal Engine farmyard capture, with 43 labelled instances outlined by NameFrameNear-nadir frame plugin_000253 of the Unreal Engine farmyard capture, with 26 labelled instances outlined by NameFrame
Map_BarnEnvironment1280×960 per frameReport previews, downscaled for the web

Download the archive.

One archive, with every modality the run wrote for these frames.

Barnyard aerial sample

25 MB

Twelve frames from a 300-frame farmyard run, shot almost straight down from 18 to 45 metres. Seven classes, 732 labelled instances, and the same reports that graded the run — including the two checks it lost points on.

Frames
12
Annotated instances
732
Format
ZIP
  • images/ rendered RGB frames, unmodified PNG
  • segmentation/ per-instance ID buffers, one colour per instance
  • labels/ YOLO boxes, one text file per frame
  • annotations/instances.json the same labels in COCO format
  • metadata/ per-frame camera, actor and spawn-manifest records
  • reports/ this capture's own report and validation output
  • capture.json the run contract: classes, camera policy, seed
  • data.yaml class names, ready for a YOLO trainer
  • TERMS.txt what you may and may not do with the pack
Download 25 MB

SHA-256f1828d704a9e9577cba8296c73e5b1ea2311d9e4b4a4d55a54e9dce5559080e6

Known limitations

The validator graded this run C · 78.3/100, and its report ships inside every archive. Here is what it and we know is wrong with the data.

  • Near-duplicate frames landed in different splits. The validator flagged it, and it is one of the two reasons the run scored 78.3. Read the report before treating any part of the shipped split as held out.
  • The gap between the commonest and the rarest class is wide enough that the validator penalised it.
  • Near-nadir at low altitude means small objects and little viewpoint variety, which is the point of the capture but also its limit.
  • Twelve packaged frames out of 300 is a sample to inspect, not a benchmark.

The checks behind those numbers, and how a run is graded, are on the dataset quality page, and the validator itself is described under dataset validation.

Terms, plainly

Free to download, read, train on and benchmark against. Publish whatever results you get, including bad ones. What you may not do is repackage the pack and sell it as your own dataset.

The images depict environment and prop content licensed for use inside Unreal Engine projects. Redistribution rights for that underlying content are not granted with these packs, so check the source licence before publishing derivatives. The full text ships as TERMS.txt inside each archive.

Citation

If you publish anything you got out of this data, this is how to point at the exact run it came from. The seed and the map are the part that matters: they are what makes the run reproducible.

NameFrame. "Barnyard: synthetic computer-vision dataset." Version 1.0, generated with Unreal Engine 5.8 on Map_BarnEnvironment, seed 20260814. https://getnameframe.com/datasets/barnyard

Need this scene, but yours?

This capture came out of one NameFrame run: a map, a class list, a camera policy and a seed. Change any of them and you get a different dataset with the same ground truth guarantees. That is what the generator is for.

How synthetic data generation works in Unreal Engine Apply for a private pilot