Minecraft Mobs Object Detection Dataset with 9 Entity Classes

Approximately 400 annotated Minecraft screenshots for object detection tasks. The dataset, created by author 'twoturtles', includes bounding box annotations for 9 entity types in COCO format. It was last updated on Hugging Face on February 9, 2026.

Use Cases

• Train object detection models based on the annotated bounding boxes for 9 Minecraft entity types.
• Benchmark model performance on synthetic imagery based on the provided Minecraft screenshots.
• Fine-tune pre-trained detectors for game-based applications based on the COCO format annotations.
• Develop real-time in-game detection systems based on the associated code mentioned in the description.

Strengths

• Annotations are provided in the standardized COCO format, which is widely supported.
• Covers 9 distinct Minecraft entity classes (chicken, cow, creeper, enderman, pig, sheep, skeleton, spider, zombie).
• Includes approximately 400 annotated images, providing a starting point for model training.

Limitations

• Row count and dataset size are unknown, which may limit suitability assessment for larger-scale projects.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset's small scale of ~400 images may limit the complexity of models it can effectively train.

Provenance

Source

Hugging Face, author twoturtles.

Collection Method

Screenshots from the Minecraft video game, annotated with bounding boxes.

Time Range

null

Freshness

Last updated 2026-02-09 23:42:23; freshness should be verified.

Geography

null