tags:
- game
- airi
- object-detection
- yolo
task_categories:
- object-detection
pretty_name: Dome Keeper YOLO Dataset v0
size_categories:
- n<1K
Dome Keeper YOLO Dataset v0
Gameplay-frame object-detection dataset for DomeKeeper YOLO v0 and the AIRI Dome Keeper plugin.
The images are already letterboxed to 640x640. Labels use normalized YOLO detection coordinates.
Splits
| Split | Images | Labeled images | Boxes |
|---|---|---|---|
| train | 226 | 214 | 269 |
| validation | 19 | 17 | 17 |
| test | 55 | 53 | 53 |
| total | 300 | 284 | 339 |
Classes
| ID | Name | Boxes |
|---|---|---|
| 0 | player_engineer |
0 |
| 1 | dome_laser |
0 |
| 2 | ore_iron |
275 |
| 3 | ore_cobalt |
24 |
| 4 | ore_water |
40 |
| 5 | enemy |
0 |
This v0 capture contains labels only for the three ore classes. The other declared classes are retained as part of the intended detector schema but have no training examples in this release.
Collection and rights
Frames and labels were generated from a manually started local Dome Keeper run by the AIRI Dome Keeper data collector. This repository does not contain decompiled game files, game packages, editor assets, or source assets extracted from the game.
Dome Keeper and its visual assets belong to their respective rights holders. This dataset is an independent research artifact and is not affiliated with or endorsed by Bippinbits or Raw Fury. Use and redistribution of gameplay imagery remain subject to the applicable game and platform terms.
References: Ultralytics detection dataset format, Dome Keeper.