datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
MoleculeNet_PCBA
MoleculeNet PCBA
PCBA (PubChem BioAssay) dataset [1], part of MoleculeNet [2] benchmark. It is intended to be used through
scikit-fingerprints library.
The task is to predict biological activity against 128 bioassays, generated by high-throughput screening (HTS). All tasks are binary active/non-active.
Note that targets have missing values. Algorithms should be evaluated only on present labels. For training data, you may want to impute them, e.g. with zeros.
Characteristic… See the full description on the dataset page: https://huggingface.co/datasets/scikit-fingerprints/MoleculeNet_PCBA.pcb-detection-augmented-seg
PCB Detection Augmented Segmentation Dataset
Augmented PCB segmentation data in YOLO format from
SanderGi/PCB-Detection.
The repository includes train, validation, and test images; segmentation labels
and masks; dataset configuration; and training outputs used by the published
PCB segmentation model.
Download
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="SanderGi/pcb-detection-augmented-seg",
repo_type="dataset"… See the full description on the dataset page: https://huggingface.co/datasets/SanderGi/pcb-detection-augmented-seg.PCB_COMPONENTS_LABELLED
PCB Components Labelled
YOLO-format object detection datasets and trained weights for detecting electronic
components on PCB (printed circuit board) images. Two dataset/model pairs are
included, covering a coarse 5-class label set and a fine-grained 17-class label set.
Contents
pcb_final_v4/ 17-class dataset (train/valid, YOLO format)
pcb_5class_tiled/ 5-class dataset, tiled images (train/valid, YOLO format)… See the full description on the dataset page: https://huggingface.co/datasets/tutitata/PCB_COMPONENTS_LABELLED.pcb-defect-segmentation
Dataset Labels
['dry_joint', 'incorrect_installation', 'pcb_damage', 'short_circuit']
Number of Images
{'valid': 25, 'train': 128, 'test': 36}
How to Use
Install datasets:
pip install datasets
Load the dataset:
from datasets import load_dataset
ds = load_dataset("keremberke/pcb-defect-segmentation", name="full")
example = ds['train'][0]
Roboflow Dataset Page
https://universe.roboflow.com/diplom-qz7q6/defects-2q87r/dataset/8… See the full description on the dataset page: https://huggingface.co/datasets/keremberke/pcb-defect-segmentation.prep_test_pcb_pile_25This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 25,
"total_frames": 17044,
"total_tasks": 1,
"total_videos": 50,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:25"
},
"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/1g0rrr/prep_test_pcb_pile_25.pcb-detection-augmented-obb
PCB Detection Augmented OBB Dataset
Augmented PCB oriented-bounding-box data in YOLO format from
SanderGi/PCB-Detection.
The repository includes train, validation, and test images; OBB labels; dataset
configuration; and training outputs used by the published PCB OBB model.
Download
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="SanderGi/pcb-detection-augmented-obb",
repo_type="dataset",
local_dir="data/augmented_obb",
)
See… See the full description on the dataset page: https://huggingface.co/datasets/SanderGi/pcb-detection-augmented-obb.PCB_defect
⚡ 데이터셋 요약
본 데이터셋은 naked PCB(부품 실장 전 PCB) 이미지로 구성되어 있으며, 총 1386장, 6종 결함, 이미지당 3~5개 결함을 포함
이미지는 실제 AOI 유사 촬영 시스템으로 템플릿 보드를 촬영한 뒤, Photoshop으로 결함을 삽입해 합성
즉, “현장(real) 결함”이 아니라, 제어된 합성 결함이라는 점이 데이터 해석/일반화에서 중요한 전제
🌳 데이터셋 구조
PCB_defect/
├── README.md
├── Meta Data/
│ └── Huang et al.(2019), A PCB Dataset for Defects Detection and Classification.pdf
├── Missing_hole/ # 결함 유형
│ ├── missing_hole01.jpg # [입력] PCB 이미지
│ ├──… See the full description on the dataset page: https://huggingface.co/datasets/RobotHuman/PCB_defect.SN-PCBAS-2026pc-benchmark
Hugging Face Resources
Resource
Type
Link
Papers Database
Dataset
ItsMaxNorm/pc-database
Papers API
Space
ItsMaxNorm/papercircle-papers-api
Benchmark Leaderboard
Space
ItsMaxNorm/pc-bench
Benchmark Results
Dataset
ItsMaxNorm/pc-benchmark
Research SessionsDataset
ItsMaxNorm/pc-research
Getting Started
Prerequisites
Node.js >= 18 and Python >= 3.10
A Supabase project
An LLM provider: Ollama (local), OpenAI, or Anthropic
Install… See the full description on the dataset page: https://huggingface.co/datasets/ItsMaxNorm/pc-benchmark.prep_test_pcb_pile_v2_16This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 16,
"total_frames": 6716,
"total_tasks": 1,
"total_videos": 32,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:16"
},
"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/1g0rrr/prep_test_pcb_pile_v2_16.Cosmos-AnomalyGen-PCB-Dataset
Dataset Overview
Dataset Description:
We are releasing 86 sample images of components on NV boards in a FoxConn factory. These images are used to finetune the AnomalyGen and Qwen-Image-Edit.
This dataset is for demonstration purposes and not for production usage.
Dataset Owner(s):
NVIDIA Corporation
Dataset Creation Date:
05/30/2026
Version:
1.0
License/Terms of Use:
GOVERNING TERMS: By downloading or… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Cosmos-AnomalyGen-PCB-Dataset.demo_grab_pcb_from_edge_igor1_pcbcrop1_aInnodisk_PCB_datasets
🚀🔬 Innodisk PCB Image Dataset
The Innodisk PCB Image Dataset is a collaboration between Innodisk Corporation and Tamkang University.Designed for research on printed‑circuit‑board (PCB) defect inspection and small‑sample learning. 🧑🔧💡
📂 Split
🖼️ Images
📁 Source directories
🏋️♂️ Train
500
7 train_* dirs
🧪 Test
240
4 test* dirs
📝 Val
30
2 val* dirs
🎯 Resolutions vary; the modal size is 188 × 128 px.
🔖 Filenames embed pseudo‑labels: pass… See the full description on the dataset page: https://huggingface.co/datasets/evan6007/Innodisk_PCB_datasets.SSP-DETR-PCB
SSP-DETR PCB Dataset
This repository contains the PCB defect-detection dataset used for the SSP-DETR experiments.
Dataset Summary
Split
Images
Annotations
Train
394
1,668
Test
99
411
The six categories are missing_hole, mouse_bite, open_circuit, short, spur, and spurious_copper.
Layout
.
|-- annotations/
| |-- instances_train.json
| `-- instances_test.json
`-- images/
|-- train/
`-- test/
Annotations follow the COCO… See the full description on the dataset page: https://huggingface.co/datasets/awesome-aning/SSP-DETR-PCB.demo_grab_pcb_from_edge_igor1_pcbcrop1This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 51,
"total_frames": 22668,
"total_tasks": 1,
"total_videos": 204,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:51"
},
"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/1g0rrr/demo_grab_pcb_from_edge_igor1_pcbcrop1.demo_grab_pcb_from_panel1_centerThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 79,
"total_frames": 40198,
"total_tasks": 1,
"total_videos": 316,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:79"
},
"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/1g0rrr/demo_grab_pcb_from_panel1_center.demo_grab_pcb_from_edge2_51demo_grab_pcb_from_panel_dag1_centerThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 51,
"total_frames": 26407,
"total_tasks": 1,
"total_videos": 204,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:51"
},
"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/1g0rrr/demo_grab_pcb_from_panel_dag1_center.test_left_grab_pcb_from_edge_50eps_cropThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 51,
"total_frames": 16209,
"total_tasks": 1,
"total_videos": 204,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:51"
},
"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/1g0rrr/test_left_grab_pcb_from_edge_50eps_crop.demo_grab_pcb_from_pile1_centerThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 52,
"total_frames": 23339,
"total_tasks": 1,
"total_videos": 208,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:52"
},
"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/1g0rrr/demo_grab_pcb_from_pile1_center.pcbslm-static-v2-unsloth-vlm
PCBSLM static-v2 Unsloth VLM
Portable multimodal Unsloth dataset for PCB layout/document-grounded training.
The JSONL splits use Unsloth/Gemma-style chat messages:
{
"messages": [
{"role": "user", "content": [
{"type": "image", "image": "assets/raw_docs/.../images/page.png"},
{"type": "text", "text": "instruction..."}
]},
{"role": "assistant", "content": [
{"type": "text", "text": "{...json answer...}"}
]}
]
}
Files… See the full description on the dataset page: https://huggingface.co/datasets/henry1477/pcbslm-static-v2-unsloth-vlm.summit_grab_pcb_from_edge_centerThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 52,
"total_frames": 19092,
"total_tasks": 1,
"total_videos": 208,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:52"
},
"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/1g0rrr/summit_grab_pcb_from_edge_center.prep_test_pcb_pile_v8_mult_all_35This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 35,
"total_frames": 10603,
"total_tasks": 1,
"total_videos": 105,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:35"
},
"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/1g0rrr/prep_test_pcb_pile_v8_mult_all_35.prep_test_flip_pcb_15This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 15,
"total_frames": 10309,
"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/1g0rrr/prep_test_flip_pcb_15.223-LeHelpingHand-PCBManipulationThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "so101_follower",
"total_episodes": 54,
"total_frames": 12006,
"total_tasks": 1,
"total_videos": 108,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:54"
},
"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/LeRobot-worldwide-hackathon/223-LeHelpingHand-PCBManipulation.lit-pcba-vs
LIT-PCBA Docked Poses (gnina virtual screening)
The docked poses from lit-pcba_vs.tgz, produced by the Koes lab at the
University of Pittsburgh for Virtual Screening with Gnina 1.0.
The original is one large archive that must be read start to finish. Here it
is split by target into lit-pcba-vs/, one archive per LIT-PCBA target, so
you can download only the targets you need:
hf download Kingldore/lit-pcba-vs lit-pcba-vs/FEN1.tar.gz --repo-type dataset
tar -xzf… See the full description on the dataset page: https://huggingface.co/datasets/Kingldore/lit-pcba-vs.prep_test_pcb_pile_v8_addsideThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 42,
"total_frames": 18972,
"total_tasks": 1,
"total_videos": 126,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:42"
},
"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/1g0rrr/prep_test_pcb_pile_v8_addside.summit_grab_pcb_from_edge_igor_dag1_cropThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 51,
"total_frames": 20404,
"total_tasks": 1,
"total_videos": 204,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:51"},
"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/1g0rrr/summit_grab_pcb_from_edge_igor_dag1_crop.PCB-AOI
PCB-AOI Dataset
Overview
PCB-AOI is an object detection dataset for Automated Optical Inspection (AOI) of Printed Circuit Boards (PCBs). It contains annotated PCB images intended for training and evaluating models that detect defects, components, or hardware constraint violations on PCBs.
The dataset is structured in YOLO format, split into train, valid, and test sets, with a data.yaml configuration file describing the classes and paths.
Dataset… See the full description on the dataset page: https://huggingface.co/datasets/kshitij1507/PCB-AOI.beta_test_pcb_pile_21This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "sam_evt2",
"total_episodes": 21,
"total_frames": 6128,
"total_tasks": 1,
"total_videos": 63,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:21"
},
"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/1g0rrr/beta_test_pcb_pile_21.