FineEdit

Note: FineEdit preview subset (1k samples) is now available. Full dataset will be released within July. Bounding Boxes This dataset provides two bounding box annotations for each sample: bbox: Obtained from an object detection tool, indicating the target region in the source image. bbox edit: Derived by computing the pixel difference between the source and edited images, followed by refinement with traditional computer vision algorithms. Citation… See the full description on the dataset page:

Source: Hugging Face Hub (zzb11/FineEdit). Metadata imported from the dataset’s Hub tags.