datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
IndustryBench-MIPU
IndustryBench-MIPU: Benchmarking Multi-Image Attribute Value Extraction for Industrial Products
Multi-Image Industrial Product Understanding Benchmark — evaluating MLLMs on structured attribute extraction from real-world industrial product images.
Industrial product specifications are scattered across multiple heterogeneous images — specification tables, nameplates, technical drawings. IndustryBench-MIPU tests whether MLLMs can reliably recover them through four… See the full description on the dataset page: https://huggingface.co/datasets/alibaba-multimodal-industrial-ai/IndustryBench-MIPU.IndustrialDetectionStaticCamerasThe IndustrialDetectionStaticCameras dataset has been collected in order to validate the methodology presented in the paper entitled A few-shot learning methodology for improving safety in industrial scenarios through universal self-supervised visual features and dense optical flow. This dataset is divided into five main folders named videoY, where Y=1,2,3,4,5. Each videoY folder contains the following:
The video of the scene in .mp4 format: videoY.mp4
A folder with the images of each frame… See the full description on the dataset page: https://huggingface.co/datasets/jjldo21/IndustrialDetectionStaticCameras.vidore_v3_industrialViDoRe V3 : Industrial reports
This dataset, Industrial reports, is a corpus of technical documents on military aircrafts (fueling, mechanics...), intended for complex-document understanding tasks. It is one of the 10 corpora comprising the ViDoRe v3 Benchmark.
About ViDoRe v3
ViDoRe V3 is our latest benchmark for RAG evaluation on visually-rich documents from real-world applications. It features 10 datasets with, in total, 26,000 pages and 3099 queries, translated into 6… See the full description on the dataset page: https://huggingface.co/datasets/vidore/vidore_v3_industrial.AgiBotWorld-Beta_G1_task_422_Packaging_items_for_industrial_logistics
agibot_task_422
This dataset converts the AgiBot format uniformly into LeRobot V3.0.
Dataset Statistics
robot_name: G1
end_effector: 夹爪
task: 为工业物流包装物品
total_episodes: 2089
total_tasks: 1
size: 150G
Dataset Structure
├── data
│ └── chunk-xxx
│ ├── file-xxx.parquet
├── meta
│ ├── episodes
│ │ └── chunk-xxx
│ │ └── file-xxx.parquet
│ ├── info.json
│ ├── stats.json
│ └── tasks.parquet
└── videos
├──… See the full description on the dataset page: https://huggingface.co/datasets/BAAI-DataCube/AgiBotWorld-Beta_G1_task_422_Packaging_items_for_industrial_logistics.vidore_v3_industrial_mteb_format
Vidore3IndustrialRetrieval
An MTEB dataset
Massive Text Embedding Benchmark
Retrieve associated pages according to questions.
Task category
t2i
Domains
Academic
Reference
https://huggingface.co/blog/QuentinJG/introducing-vidore-v3
Source datasets:
vidore/vidore_v3_industrial
How to evaluate on this task
You can evaluate an embedding model on this dataset using the following code:
import mteb
task = mteb.get_task("Vidore3IndustrialRetrieval")… See the full description on the dataset page: https://huggingface.co/datasets/vidore/vidore_v3_industrial_mteb_format.industrial_cart_2axis-ego-centric-data-industrial-samplesindustrial-instruction-dataset
Industrial-Instruction Dataset
Industrial-Instruction provides benchmark and training-ready QA instances derived from industrial technical reports, designed to evaluate robustness under realistic retrieval conditions. Samples are grounded in retrieved evidence and include irrelevant retrieval, single-/multi-document support, and single-/multi-document answer settings.
Paper
Industrial-Instruction: An End-to-End Framework for Building Instruction-Tuning and… See the full description on the dataset page: https://huggingface.co/datasets/Parssky/industrial-instruction-dataset.XMR_Industrial_Foreign_Object_Detection_Lentils
Hyperspectral Foreign-Object Detection in Lentils — Full Dataset
The larger counterpart to the small tutorial demo at
cubert-gmbh/XMR_Demo_Industrial_Foreign_Object_Detection_Lentils.
Captured with a Cubert Ultris XMR camera — 61 bands per pixel, 430–910 nm, 1080 × 1000 pixels. Three acquisition days, 15 merged .cu3s capture sessions, 1,136 frames total, 696 frames with pixel-level COCO annotations across 7 foreign-object classes.
Foreign-object detection in… See the full description on the dataset page: https://huggingface.co/datasets/cubert-gmbh/XMR_Industrial_Foreign_Object_Detection_Lentils.industrial-semseg-web
Industrial 3D Semantic Segmentation — Web Display Assets
Display assets for the dataset website of a synthetic industrial RGB-D dataset for 3D
semantic segmentation, generated with NVIDIA Isaac Sim. Files in this repository back
the static website and are meant to be fetched directly via resolve URLs; the full
datasets and model checkpoints live in separate repositories.
Quantitative benchmark results are not included here; they will be published together
with the thesis.… See the full description on the dataset page: https://huggingface.co/datasets/min99ian/industrial-semseg-web.X4_SWIR_Industrial_Foreign_Object_Detection_Bedding
Hyperspectral Foreign-Object Detection in Bedding — Full Dataset (VIS + SWIR)
A 6-band VIS+SWIR hyperspectral dataset for industrial foreign-object
detection on a bedding substrate (a tray of wood-shaving / sawdust animal
bedding). Captured with a Cubert Ultris X4 + SWIR rig — 6 spectral bands
at 450 / 550 / 625 nm (VIS) and 1050 / 1200 / 1450 nm (SWIR), 2400 × 4900
pixels per frame. 252 frames (193 train · 59 val), 51 frames carry
pixel-level polygon… See the full description on the dataset page: https://huggingface.co/datasets/cubert-gmbh/X4_SWIR_Industrial_Foreign_Object_Detection_Bedding.industrial-defects-detection
Industrial Defects Detection Dataset
Professional industrial defects detection dataset for AI/ML training.
Sample Dataset (60 images)
This repository contains a free sample of 60 images across 6 categories.
Full Dataset
Package
Images
Resolution
Price
Standard
2,460
512x512
15 EUR
Premium
2,460
2048x2048
Coming soon
Buy Full Dataset on Gumroad: https://betosfyro.gumroad.com/l/mumbyd
Categories
corrosion, crack, dent… See the full description on the dataset page: https://huggingface.co/datasets/xanoutas/industrial-defects-detection.IndustrialTextileDataset
Description
Introduction of new dataset for unsupervised fabric defect detection
This dataset aims to provide a color dataset with real industrial fabric defect gathered in a visiting machine with several industrial cameras.
It has been designed with the same nomenclature as MVTEC AD dataset (https://www.mvtec.com/company/research/datasets/mvtec-ad) for unsupervised anomaly detection.
Type
Total
Train(Good)
Test(Good)
Test(Defective)
Sample
type1cam1386… See the full description on the dataset page: https://huggingface.co/datasets/SimTho/IndustrialTextileDataset.Industrial-Site-Safety-Detection-v1-DATASETA dataset for yolo built for the Industrial-Site-Safety-Detection-v1 model. Data was scraped from various youtube operating videos. Dataset consists of 415 labels and images.
Classes are: backhoe_loader
bucket
bucket_empty
bucket_full
bulldozer
dump_truck
excavator
grader
person_no_ppe
person_ppe
road_roller
safety_barrel
safety_cone
trench
wheel_loader
industrial_cartrocochallenge2026_Industrial_Assembly_annotatedThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 10,
"features": {
"observation.images.head": {
"dtype": "video",
"shape": [
240,
320,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height":… See the full description on the dataset page: https://huggingface.co/datasets/nikodembartnik/rocochallenge2026_Industrial_Assembly_annotated.sangyo_no_yume_industrial_dreams
From the Frontier Research Team at Takara.ai we present the "Sangyo no Yume Industrial Dreams" dataset, a collection of AI-generated industrial dreamscapes.
Sangyo no Yume Industrial Dreams
Dataset Details
"Sangyo no Yume Industrial Dreams" is a collection of images generated using SDXL Lightning with specialized prompt engineering techniques. These images balance industrial themes with dreamlike qualities, creating a unique aesthetic that sits at the intersection of… See the full description on the dataset page: https://huggingface.co/datasets/takara-ai/sangyo_no_yume_industrial_dreams.Industrial-Workplace-Egocentric-FHD-Samples
Industrial & Workplace Egocentric Video — FHD Samples
21 clips of first-person video of real industrial and workplace tasks, captured with a head-mounted smartphone. Released by TrainThemAI for training Vision-Language-Action (VLA) models, World Action Models (WAM), and humanoid manipulation policies — π0, π1, OpenVLA, RT-2, GR00T, Cosmos, DreamZero.
Fully rights-cleared, MIT-licensed, and representative of our production capture pipeline.
Companion to POV Egocentric Video —… See the full description on the dataset page: https://huggingface.co/datasets/TrainThemAI/Industrial-Workplace-Egocentric-FHD-Samples.industrial-harbor-assetsindustrial-parts-and-packaging
Industrial Parts & Packaging
Try it before you download. Every item in this dataset is live in
projectsim Lab: orbit every variant at real scale, then grab the
full USD (physics, PBR maps) or a GLB preview per asset.
Jump straight to a class: Gas Cylinders · Pallet Racking.
Procedurally generated, sim-ready 3D industrial infrastructure - warehouse
pallet racking bays and LPG / propane gas cylinders - shipped as OpenUSD
(.usda) with per-zone PBR maps, plus a static… See the full description on the dataset page: https://huggingface.co/datasets/projectsim/industrial-parts-and-packaging.industrial_manipulation_egocentric_sample
Industrial Manipulation — Egocentric Video (Sample)
Head-mounted video of industrial manipulation tasks, recorded by workers during real production shifts. Deformable materials and metal handling.
This is a 16-clip sample from an ongoing collection. It exists so you can judge the footage for yourself and tell us whether it's useful.
What's here
Clips
16
Total duration
8 minute 14 seconds
Camera
Head-mounted, DJI Osmo Action 5 Pro
Resolution… See the full description on the dataset page: https://huggingface.co/datasets/sovrano-ai/industrial_manipulation_egocentric_sample.Industrial_testindustrial_screws_sorting_expertThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.0",
"robot_type": "so100",
"total_episodes": 20,
"total_frames": 29859,
"total_tasks": 1,
"total_videos": 40,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:20"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/kantine/industrial_screws_sorting_expert.industrial-technical-archive
🚀 Latest Updates (July, 2026)
Version: v07.2026 (Verified)
Status: Integrated with 1,000,000+ records.
New Files: product-E-26-07-2026.csv & product-V-26-07-2026.csv.
QTE Technologies: Industrial & Scientific Knowledge Base
Wikidata Entity: Q138411149
IPFS CID: bafybeibogxxuhmzfrsuhcfd4qr4tmc4okhmrcwhp3266hq47ccuyjnjxoq
Official Neural Hub: qtetech.github.io
This is the permanent technical archive for QTE Technologies, ensuring long-term accessibility of… See the full description on the dataset page: https://huggingface.co/datasets/QTE-Technologies/industrial-technical-archive.IndustryBench
IndustryBench: Probing the Industrial Knowledge Boundaries of LLMs
💻Github | 📝Paper
IndustryBench is a multi-lingual benchmark for evaluating the industrial domain knowledge of large language models. It comprises 2,049 expert-curated QA pairs spanning 12 industrial sectors, with human-reviewed translations in Chinese, English, Russian, and Vietnamese.
Overview
Dimension
Details
Total questions
2,049
Languages
Chinese (zh), English (en), Russian (ru)… See the full description on the dataset page: https://huggingface.co/datasets/alibaba-multimodal-industrial-ai/IndustryBench.industrial-maintenance-synthetic
Industrial Maintenance Synthetic Dataset
Synthetic dataset of 1M sensor tag records and 1M maintenance work orders across 37 industrial equipment types, generated for domain-specific NLP models in industrial maintenance.
Dataset Details
Property
Value
Total rows
~2M (1M sensor tags + 1M maintenance records)
Equipment types
37
Equipment instances
145
Languages
English
Format
Parquet
With impurities
Yes (10% rate)
Build date
2026-03-07
License… See the full description on the dataset page: https://huggingface.co/datasets/Jvachier/industrial-maintenance-synthetic.XMR_Demo_Industrial_Foreign_Object_Detection_Lentils
Demo for Hyperspectral Foreign-Object Detection in Lentils
Video spectroscopy beyond the visible spectrum, applied to foreign-object detection on a sliding lentil conveyor. Captured with a Cubert Ultris XMR camera — 61 bands per pixel, 430–910 nm, 1080 × 1000 pixels at 4 fps.
Foreign-object detection in food sorting is a general industrial-inspection problem — the rejected target could be a stone, a stem, a piece of packaging, a metal shard, or an insect. In this… See the full description on the dataset page: https://huggingface.co/datasets/cubert-gmbh/XMR_Demo_Industrial_Foreign_Object_Detection_Lentils.industrial_soldering_anomalyThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.0",
"robot_type": "so100",
"total_episodes": 15,
"total_frames": 13440,
"total_tasks": 1,
"total_videos": 30,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:15"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/kantine/industrial_soldering_anomaly.LongTS-Industrial
LongTS-Industrial Benchmark
面向视觉语言模型（VLM）长时序工业故障推理能力的评测基准。将长时间序列的多通道工业监测数据可视化为图像，要求模型通过「看图」完成从整体感知、异常定位、根因分析到运维决策的分层推理。
💻 代码（数据合成 / 基准构建 / 评测）：https://github.com/ctacyk/MERIT
📄 论文：（论文发表后补充）
数据概览
样本数：1716（4 个设备域，143 个场景，序列长度 ∈ {1000, 3000, 5000, 10000}，随机种子 ∈ {42, 123, 789}）
设备域：磨煤机 coal_mill（576）、水泵 pump（372）、变压器 transformer（372）、风电 wind_turbine（396）
生成方式：全部由 TimeBlender 物理建模合成（热惯性 + 通道耦合 + 根因传播的故障注入），并以真实工业数据统计特征标定，非真实采集数据。
任务分层（L1–L4）… See the full description on the dataset page: https://huggingface.co/datasets/ANTICH/LongTS-Industrial.industrial_screws_sorting_anomalyThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.0",
"robot_type": "so100",
"total_episodes": 10,
"total_frames": 14926,
"total_tasks": 1,
"total_videos": 20,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:10"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/kantine/industrial_screws_sorting_anomaly.