CoolFace
Datasetpublic

ynyg/InkEraser

ynyg/InkEraser 筆跡擦除數據集,用於將含有筆跡的圖像(source)轉換為去除筆跡的目標圖像(target)。 文件結構 data/ train/ source/ # 原圖(含筆跡) target/ # 目標(去筆跡) val/ source/ target/ 配對規則 source/ 與 target/ 中使用完全相同的文件名進行一一對應。 圖像格式:.jpg。 數據量(當前倉庫) train:1511 對 val:115 對 合計:1626 對 使用示例(Python) from huggingface_hub import snapshot_download from pathlib import Path from PIL import Image root = Path(snapshot_download("ynyg/InkEraser"… See the full description on the dataset page: https://huggingface.co/datasets/ynyg/InkEraser.

sourceHugging Faceapache-2.0updated 8mo agoView on Hugging Face
0likes179downloads
Dataset Card

ynyg/InkEraser

筆跡擦除數據集,用於將含有筆跡的圖像(source)轉換為去除筆跡的目標圖像(target)。

文件結構

data/
  train/
    source/   # 原圖(含筆跡)
    target/   # 目標(去筆跡)
  val/
    source/
    target/

配對規則

  • source/target/ 中使用完全相同的文件名進行一一對應。
  • 圖像格式:.jpg

數據量(當前倉庫)

  • train:1511 對
  • val:115 對
  • 合計:1626 對

使用示例(Python)

python
from huggingface_hub import snapshot_download
from pathlib import Path
from PIL import Image

root = Path(snapshot_download("ynyg/InkEraser", repo_type="dataset"))

for split in ["train", "val"]:
    src_dir = root / "data" / split / "source"
    tgt_dir = root / "data" / split / "target"

    for src_path in src_dir.glob("*.jpg"):
        tgt_path = tgt_dir / src_path.name
        # 成對讀取
        src_img = Image.open(src_path)
        tgt_img = Image.open(tgt_path)
        # TODO: your training code

授權

Apache-2.0