README.md

metadata

license: apache-2.0
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
dataset_info:
  features:
    - name: image
      dtype: image
    - name: label
      dtype:
        class_label:
          names:
            '0': Safe
            '1': Unsafe
  splits:
    - name: train
      num_bytes: 10844205860
      num_examples: 79997
  download_size: 12053208752
  dataset_size: 10844205860
task_categories:
  - image-classification
language:
  - en
tags:
  - ImageShield
  - safe
  - unsafe
  - Guardrail
pretty_name: ImageShield
size_categories:
  - 10K<n<100K

This dataset contains sensitive visual content intended for training Image Guard, Image Filter, and Image Shield models. Due to the sensitive nature of the data, access is gated and requires manual review. Access may be approved for users who provide a valid research, safety, or development purpose.

ImageShield-Guardrail-80K

ImageShield-Guardrail-80K is a curated dataset containing ~80,000 labeled images designed for training, benchmarking, and fine-tuning computer vision safety guardrails, content moderation filters, and image protection models (such as ImageShield, Image Guard, and automated content screening systems).

• Curator: prithivMLmods
• Total Samples: 79,997 rows
• Total Size: ~12.1 GB
• Format: Parquet / Image Feature
• Split: Train (80k rows)

Dataset Overview

The dataset provides binary annotations to classify visual content into Safe vs. Unsafe categories, with dedicated coverage for sensitive visual content, inappropriate adult content, and general AI safety and content moderation guardrail enforcement.

Column	Type	Description
image	Image	Target RGB image (varying resolutions)
label	ClassLabel	0 = Safe, 1 = Unsafe

Access & Gating Notice

Sensitive Content Warning: This dataset contains sensitive visual content and safety-focused training representations intended strictly for research, content moderation, and safety guardrail engineering. Access is gated and subject to manual review.

To gain access:

1. Log in to your Hugging Face account.
2. Submit an access request outlining your intended research, safety filtering, or development use case.
3. Access will be approved upon review.

How to Use

Using datasets

from datasets import load_dataset

# Load dataset (requires approved access token)
dataset = load_dataset("prithivMLmods/ImageShield-Guardrail-80K", split="train")

# Access a single sample
sample = dataset[0]
image = sample["image"]
label = sample["label"]

print(f"Label: {'Unsafe' if label == 1 else 'Safe'}")

PyTorch DataLoader Example

from datasets import load_dataset
from torchvision import transforms
from torch.utils.data import DataLoader

transform = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
])

def preprocess(batch):
    batch["pixel_values"] = [transform(img.convert("RGB")) for img in batch["image"]]
    return batch

dataset = load_dataset("prithivMLmods/ImageShield-Guardrail-80K", split="train")
dataset = dataset.with_transform(preprocess)

dataloader = DataLoader(dataset, batch_size=32, shuffle=True)

Intended Uses

• Training lightweight edge guardrail classifiers (e.g., MobileNet, EfficientNet, FastViT).
• Safety alignment and pre-generation/post-generation filtering for Vision-Language Models (VLMs) and diffusion pipelines.
• Automated moderation and policy enforcement benchmarking.

License

This dataset is released under the Apache-2.0 License. Users must adhere to safety and ethical AI guidelines, ensuring models developed from this data are used exclusively for defense, moderation, and content safety systems.