convert_to_imagefolder.py

importargparse
importjson
importpathlib


parser=argparse.ArgumentParser()
parser.add_argument(
"--path",
type=str,
required=True,
help="Path to folder with image-text pairs.",
)
parser.add_argument("--caption_column", type=str, default="prompt", help="Name of caption column.")
args=parser.parse_args()

path=pathlib.Path(args.path)
ifnotpath.exists():
raiseRuntimeError(f"`--path` '{args.path}' does not exist.")

all_files=list(path.glob("*"))
captions=list(path.glob("*.txt"))
images=set(all_files) -set(captions)
images= {image.stem: imageforimageinimages}
caption_image= {caption: images.get(caption.stem) forcaptionincaptionsifimages.get(caption.stem)}

metadata=path.joinpath("metadata.jsonl")

withmetadata.open("w", encoding="utf-8") asf:
forcaption, imageincaption_image.items():
caption_text=caption.read_text(encoding="utf-8")
json.dump({"file_name": image.name, args.caption_column: caption_text}, f)
f.write("\n")