You can deploy this model and start earning money today!

aimagelab/
ViSU-Text

Views

No views yet

ViSU (Visual Safe and Unsafe) Dataset

Welcome to the official page of the ViSU (Visual Safe and Unsafe) dataset, introduced for the first time in Safe-CLIP paper. This README provides an overview of the dataset and instructions on how to use it.

Dataset Overview

The ViSU dataset is composed of quadruplets, each consisting of:

A safe sentence (from COCO). A corresponding safe image (from COCO). An NSFW sentence semantically correlated with the safe one (AI-generated). A… See the full description on the dataset page: https://huggingface.co/datasets/aimagelab/ViSU-Text.