RealText-V1

RealText-V1: A Text-Centric Image Forgery Analysis Dataset 💾 Dataset Description RealText-V1 is a text-centric image forgery analysis dataset built to benchmark visual-logical co-reasoning over text-centric image forgeries. It pairs forged and pristine document-like text images with pixel-level manipulation masks and expert-level natural-language explanations that ground every verdict in observable visual and logical evidence. RealText-V1 is the dataset… See the full description on the dataset page:

Source: Hugging Face Hub (vankey/RealText-V1). Metadata imported from the dataset’s Hub tags.