datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
MET-Bench-Minecraft
MET-Bench: Multimodal Entity Tracking for Evaluating the Limitations of Vision-Language and Reasoning Models
Vanya Cohen and Raymond Mooney · ICML 2026
Paper · Publication page · Load the dataset · Citation
Domains: Chess · Shell Game · Minecraft
MET-Bench evaluates entity state tracking across text and image modalities. This repository contains the Minecraft domain.
Minecraft
Minecraft is a state prediction task involving partial observations, dynamic… See the full description on the dataset page: https://huggingface.co/datasets/vanyacohen/MET-Bench-Minecraft.MET-Bench-Minecraft-Trajectories
MET-Bench: Multimodal Entity Tracking for Evaluating the Limitations of Vision-Language and Reasoning Models
Vanya Cohen and Raymond Mooney · ICML 2026
Paper · Evaluation code · Minecraft benchmark · Usage
Benchmark domains: Chess · Shell Game · Minecraft
Minecraft trajectories
This dataset contains the 462 source recordings used to construct the released MET-Bench Minecraft benchmark, comprising 462,235 captured observations. The recordings follow scripted… See the full description on the dataset page: https://huggingface.co/datasets/vanyacohen/MET-Bench-Minecraft-Trajectories.Minecraft-Skins-20M
Dataset Card for Minecraft Skins
Dataset Summary
This dataset contains 19,973,928 unique Minecraft player skins collected from various sources. Each skin is stored as a base64-encoded image with a unique identifier.
Dataset Structure
Data Fields
This dataset includes the following fields:
id: A randomly generated UUID for each skin entry. These UUIDs are not linked to any external APIs or services (such as Mojang's player UUIDs) and serve solely as… See the full description on the dataset page: https://huggingface.co/datasets/nyuuzyou/Minecraft-Skins-20M.minecraft-skins-1.1m-prerenderedMinecraft-Skins-Captioned-1M
Dataset Card for Minecraft Skins
Dataset Summary
This dataset contains 981,079 unique Minecraft player skins collected from various sources. Each skin is stored as a base64-encoded image with a unique identifier.
Dataset Structure
Data Fields
This dataset includes the following fields:
hash: A data dependent hash. These hashes are generated from raw bytes and will be same if the skin is identical.
image: The skin image encoded in base64 format.… See the full description on the dataset page: https://huggingface.co/datasets/neurlang/Minecraft-Skins-Captioned-1M.minecraft-skins-captioned-900k
🎮 minecraft-skins-captioned-900k
854,116 high-quality, captioned Minecraft player skins — deduplicated, Steve-model only, ready for text-to-image training.
📋 Dataset Summary
A rigorously filtered and quality-controlled version of neurlang/Minecraft-Skins-Captioned-1M specifically curated for training high-performance generative models that require precise UV topology constraints.
This dataset is optimized for models like ST-DiT (Sparse Template-Aware… See the full description on the dataset page: https://huggingface.co/datasets/summykai/minecraft-skins-captioned-900k.Minecraft-winter-shaders-Lite
❄️ Minecraft Winter Shaders (Lite)
10,000+ winter Minecraft screenshots with Photon Shaders.
For generative models, style transfer, and high-quality vision tasks.
Structure
Single folder: winter/
Format: 960x540 JPEG
License
Dataset: CC BY NC 4.0.Shader: Photon Shaders by SixthSurge (commercial use of screenshots explicitly allowed).See PHOTON_SHADERS_LICENSE.txt.
Minecraft-vision-dataset
🌍 Minecraft Vision Dataset
Manually collected Minecraft screenshots for classification tasks.
Useful for training AI agents, bots, and computer vision models.
📦 What's inside
Biomes, dimensions, time of day, and more.
See the repository file browser for the full list of classes and current image counts.
📂 Dataset structure
The root contains one folder per class. New classes are added over time.
🔍 Current statistics
Exact numbers… See the full description on the dataset page: https://huggingface.co/datasets/Nininkkka/Minecraft-vision-dataset.CraftSight-Minecraft
CraftSight
Frame-level multi-label visual annotations for Minecraft agents.
Dataset creation toolkit: CraftSight is built and maintained with CraftSight Labeler, an open-source, browser-based annotation tool and reproducible release pipeline for multi-label Minecraft vision datasets. It supports manual and model-assisted labeling, structured game-state annotations, and trajectory-safe train/validation/test splits.
CraftSight provides Minecraft gameplay frames annotated… See the full description on the dataset page: https://huggingface.co/datasets/Krows7/CraftSight-Minecraft.minecraft-preview
Overview
The Minecraft Character Dataset was used to fine-tune the Stable Diffusion Minecraft Character Preview model.
It currently consists of 1022 images of forward facing and rear facing 3D renders of various Minecraft character skins.
Contact Information
You can contact me at: Cory Spencer <cory@monadical.com>
minecraft-biomes
Minecraft Biomes (RGBD, pseudo-labeled)
Pseudo-labeled RGBD screenshots from Minecraft, covering 12 broad biome
categories. Each sample is an (RGB, depth) pair at 640×360 resolution.
Source
RGBD frames: 1908 paired (rgb, depth) samples from
zid8/syntheticMinecraftRGBD,
collected via MineRL.
Labels: Generated by a Gemma-3-4B model fine-tuned with LoRA on
the willowc/minecraft-biomes
dataset, then augmented with ~160 hand-selected MineRL ocean frames
to fix an… See the full description on the dataset page: https://huggingface.co/datasets/Wafik20/minecraft-biomes.minecraft-skins
minecraft-skins
A curated, high-quality dataset of Minecraft character skins with descriptive captions and tags for training generative AI diffusion models (such as Stable Diffusion 1.5, SDXL, and FLUX LoRAs).
Dataset Summary
Total Skins: 101
Image Resolution: 64x64 RGBA (standard Minecraft 1.8+ dual-layer UV layout)
Model Variants: Classic (4px arms) and Slim (3px arms)
Metadata Fields: image, username, uuid, caption, tags, model, quality
Quick… See the full description on the dataset page: https://huggingface.co/datasets/honvert/minecraft-skins.minecraft-mobs-yolo-dataset
🎮 Minecraft Mobs YOLO Dataset
2,585+ pre-split images with YOLO bounding box annotations for
Minecraft mob detection across 5 macro-classes.
🧭 Overview
This dataset contains 2,585 images of Minecraft mobs annotated
with bounding boxes in YOLO format. Pre-split into Train (80%)
and Validation (20%) sets — ready to plug into any
Ultralytics YOLO pipeline.
🧟 Classes (5 Macro-Classes)
ID
Class
Instances
Includes
0
creeper
594
Standard… See the full description on the dataset page: https://huggingface.co/datasets/hmnshudhmn24/minecraft-mobs-yolo-dataset.Minecraft-Fable-Renders-v2minecraft-captioningMinecraft-Fable-ImageGLB-v1minecraft-skins-20k-1024k-captionedminecraft-mobsMinecraft mob object detection
~400 Minecraft screenshots annotated with bounding boxes for 9 entity types.
COCO format annotations.
Classes
0: chicken, 1: cow, 2: creeper, 3: enderman, 4: pig, 5: sheep, 6: skeleton, 7: spider, 8: zombie
See Also
model
code
The code allows you to run the model over Minecraft in realtime.
Sample Detection
spatial-minecraftminecraft-skins-1.1m-deduped-64x64
Minecraft Skins 1.1M Deduped (64x64 Edition)!
Nyuuzyou's Minecraft-Skins-20M but deduped using BLAKE3 hashes (only catches if pixel values are exactly the same), then filtered to only valid 64x64 skins.
Format is a 2.6 GB ZIP archive containing 64x64 PNG skin files.
Tools used
PIL Image (Python), Google Colab (free CPU tier) and BLAKE3 (Python)
How it was made
Loaded Nyuuzyou's Minecraft-Skins-20M,
Deduped using BLAKE3, only catching if pixel… See the full description on the dataset page: https://huggingface.co/datasets/MihaiPopa-1/minecraft-skins-1.1m-deduped-64x64.minecraft-crafting-vqa-ru-en
Minecraft Crafting & Gameplay VQA Dataset (RU/EN)
A bilingual, multimodal dataset specifically designed for fine-tuning Vision-Language Models (VLMs) such as Qwen2.5-VL and Qwen3-VL. This dataset is optimized for training AI models to recognize crafting recipes, understand in-game user interfaces, and trigger function calls (Tool Calling).
Repository Structure
The dataset is cleanly separated into language locales. Each folder contains its own data.json annotation file… See the full description on the dataset page: https://huggingface.co/datasets/KuroTo4ka/minecraft-crafting-vqa-ru-en.minecraft-skins-100k-train-readyMinecraft-16x-DatasetMinecraft-Depth-Images
D4MCDataset
Dataset created to train the Depth4MC model.
Images and depth label are of size 480x854.
Read more information on the GitHub page: https://github.com/JulianBvW/Depth4MC
MinecraftRigsminecraft-skins-1.1m-deduped-64x64-1.5
Minecraft Skins 1.1M Deduped (64x64 Edition) 1.5!
Minecraft Skins 1.1M Deduped but troll skins were filtered (more this time).
Format is a 2.5 GB ZIP archive containing 64x64 PNG skin files.
Tools used
PIL Image (Python) and Google Colab (free CPU tier)
How it was made
Loaded Minecraft Skins 1.1M Deduped,
Troll (like single-color) skins were filtered (std > 10 were kept) and removed.
Result is 1100250 real 64x64 skins.
Output is given in a 2.5 GB… See the full description on the dataset page: https://huggingface.co/datasets/MihaiPopa-1/minecraft-skins-1.1m-deduped-64x64-1.5.minecraft-sampleminecraft-skinminecraft-skins-legacy
Minecraft Skin Img2Img Dataset (Legacy Format)
Data Overview
[!NOTE]
This dataset contains skins in the pre-1.8 format, with an image size of 64px x 32px instead of the newer 64px x 64px.
This dataset contains character images paired with a Minecraft skin texture and their descriptions. Character images were generated by DALL-E, prompted with renderings of the Minecraft skin file. All images in this dataset are 1024px x 1024px.
Usage
You can load this dataset… See the full description on the dataset page: https://huggingface.co/datasets/deepwaterhorizon/minecraft-skins-legacy.minecraft_items