update README.md

README.md

@@ -11,22 +11,79 @@ size_categories: 1K<n<10K
 
 # Wireframe Dataset
 
-This is the **Wireframe dataset** hosted on Hugging Face Hub.
+This is the [**Wireframe dataset**](https://github.com/huangkuns/wireframe) hosted on Hugging Face Hub.
 
-## Dataset Summary
+## Summary
 
 Wireframe dataset with image annotations including line segments.  
-The dataset is stored as pickle files (`train.pkl`, `test.pkl`) and images.  
-A custom `dataset.py` is provided for loading.
+The dataset is stored as jsonl files (`train/metadata.jsonl`, `test/metadata.jsonl`) and images.
+
+## Download
+
+- Download with huggingface-hub
+
+```shell
+python3 -m pip install huggingface-hub
+huggingface-cli download --repo-type dataset lh9171338/Wireframe --local-dir ./
+```
+
+- Download with Git
+
+```shell
+git lfs install
+git clone https://huggingface.co/datasets/lh9171338/Wireframe
+```
 
 ## Usage
 
+- Load the dataset from Hugging Face Hub
+
 ```python
 from datasets import load_dataset
 
-# Load the dataset from Hugging Face Hub
 ds = load_dataset("lh9171338/Wireframe")
+print(ds)
+# DatasetDict({
+#     train: Dataset({
+#         features: ['image', 'image_file', 'image_size', 'lines'],
+#         num_rows: 5000
+#     })
+#     test: Dataset({
+#         features: ['image', 'image_file', 'image_size', 'lines'],
+#         num_rows: 462
+#     })
+# })
+print(ds["train"][0].keys())
+# dict_keys(['image', 'image_file', 'image_size', 'lines'])
+```
 
-# Access samples
+- Load the dataset from local
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("imagefolder", data_dir=".")
 print(ds)
-print(ds["train"][0])
+# DatasetDict({
+#     train: Dataset({
+#         features: ['image', 'image_file', 'image_size', 'lines'],
+#         num_rows: 5000
+#     })
+#     test: Dataset({
+#         features: ['image', 'image_file', 'image_size', 'lines'],
+#         num_rows: 462
+#     })
+# })
+print(ds["train"][0].keys())
+# dict_keys(['image', 'image_file', 'image_size', 'lines'])
+```
+
+- Load the dataset with jsonl files
+```python
+import jsonlines
+
+with jsonlines.open("train/metadata.jsonl") as reader:
+    infos = list(reader)
+print(infos[0].keys())
+# dict_keys(['file_name', 'image_file', 'image_size', 'lines'])
+```

dataset.py

@@ -1,90 +0,0 @@
-# -*- encoding: utf-8 -*-
-
-"""
-@File    :   dataset.py
-@Time    :   2025/08/31 23:00:00
-@Author  :   lh9171338
-@Version :   1.0
-@Contact :   2909171338@qq.com
-"""
-
-import os
-import tqdm
-import pickle
-import jsonlines
-from datasets import Dataset, Features, Sequence, Value
-
-
-def convert_pkl_to_parquet(split):
-    """
-    Convert pkl file to parquet file
-
-    Args:
-        split (str): split name
-
-    Returns:
-        None
-    """
-    # 加载 pkl 文件
-    with open(f"{split}/{split}.pkl", "rb") as f:
-        annotations = pickle.load(f)
-
-    # 新增 file_name 字段
-    new_annotations = []
-    for ann in tqdm.tqdm(annotations):
-        new_ann = dict(
-            file_name=ann["image_file"],
-            image_file=os.path.basename(ann["image_file"]),
-            image_size=ann["image_size"],
-            lines=ann["lines"].tolist(),
-        )
-        new_annotations.append(new_ann)
-
-    # 转 Dataset
-    features = Features(
-        {
-            "file_name": Value("string"),
-            "image_file": Value("string"),
-            "image_size": Sequence(Value("int32")),
-            "lines": Sequence(Sequence(Sequence(Value("float")))),
-        }
-    )
-    ds = Dataset.from_list(new_annotations, features=features)
-
-    # 保存 parquet
-    ds.to_parquet(f"{split}/metadata.parquet")
-
-
-def convert_pkl_to_jsonl(split):
-    """
-    Convert pkl file to jsonl file
-
-    Args:
-        split (str): split name
-
-    Returns:
-        None
-    """
-    # 加载 pkl 文件
-    with open(f"{split}/{split}.pkl", "rb") as f:
-        annotations = pickle.load(f)
-
-    # 新增 file_name 字段
-    new_annotations = []
-    for ann in tqdm.tqdm(annotations):
-        new_ann = dict(
-            file_name=ann["image_file"],
-            image_file=os.path.basename(ann["image_file"]),
-            image_size=ann["image_size"],
-            lines=ann["lines"].tolist(),
-        )
-        new_annotations.append(new_ann)
-
-    # 保存 jsonl文件
-    with jsonlines.open(f"{split}/metadata.jsonl", "w") as f:
-        f.write_all(new_annotations)
-
-
-if __name__ == "__main__":
-    convert_pkl_to_jsonl("train")
-    convert_pkl_to_jsonl("test")

test/test.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:098be958c6eb01c29ff781c3a6ea3433326aa2ddc24faa54aa5c9797d143a0df
-size 584354

train/train.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:56b5d251e90eb925b0ac8414eb21907ccc00cb642b7214269f518458d8bc5aa1
-size 6383995