Dataset Viewer
The dataset viewer is not available for this dataset.
Cannot get the config names for the dataset.
Error code:   ConfigNamesError
Exception:    ValueError
Message:      Feature type 'BBoxFeature' not found. Available feature types: ['Value', 'ClassLabel', 'Translation', 'TranslationVariableLanguages', 'LargeList', 'List', 'Array2D', 'Array3D', 'Array4D', 'Array5D', 'Audio', 'Image', 'Mesh', 'Video', 'Pdf', 'Nifti', 'Json']
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/dataset/config_names.py", line 67, in compute_config_names_response
                  config_names = get_dataset_config_names(
                      path=dataset,
                      token=hf_token,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 161, in get_dataset_config_names
                  dataset_module = dataset_module_factory(
                      path,
                  ...<4 lines>...
                      **download_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/load.py", line 1215, in dataset_module_factory
                  raise e1 from None
                File "/usr/local/lib/python3.14/site-packages/datasets/load.py", line 1190, in dataset_module_factory
                  ).get_module()
                    ~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/load.py", line 700, in get_module
                  config_name: DatasetInfo.from_dict(dataset_info_dict)
                               ~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/info.py", line 284, in from_dict
                  return cls(**{k: v for k, v in dataset_info_dict.items() if k in field_names})
                File "<string>", line 20, in __init__
                File "/usr/local/lib/python3.14/site-packages/datasets/info.py", line 170, in __post_init__
                  self.features = Features.from_dict(self.features)
                                  ~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/features/features.py", line 2009, in from_dict
                  obj = generate_from_dict(dic)
                File "/usr/local/lib/python3.14/site-packages/datasets/features/features.py", line 1588, in generate_from_dict
                  return {key: generate_from_dict(value) for key, value in obj.items()}
                               ~~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/features/features.py", line 1604, in generate_from_dict
                  return Sequence(feature=generate_from_dict(feature), **obj)
                                          ~~~~~~~~~~~~~~~~~~^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/features/features.py", line 1588, in generate_from_dict
                  return {key: generate_from_dict(value) for key, value in obj.items()}
                               ~~~~~~~~~~~~~~~~~~^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/features/features.py", line 1594, in generate_from_dict
                  raise ValueError(f"Feature type '{_type}' not found. Available feature types: {list(_FEATURE_TYPES.keys())}")
              ValueError: Feature type 'BBoxFeature' not found. Available feature types: ['Value', 'ClassLabel', 'Translation', 'TranslationVariableLanguages', 'LargeList', 'List', 'Array2D', 'Array3D', 'Array4D', 'Array5D', 'Audio', 'Image', 'Mesh', 'Video', 'Pdf', 'Nifti', 'Json']

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

MNIST 合成数字数据集(640×640) / Synthetic MNIST Digits Dataset

由本项目 YOLOv5 合成器(data/scripts/make_digit_images.py全合成的数字(0–9)数据集: 多字体矢量渲染 + MNIST 手写体放大着色双产线;模式为 multi(检测向,每图 2–5 目标 + 形近干扰字符)。 所有图像均为本项目生成,MNIST 原始素材仅作为渲染数据源(源级零泄漏:train 只用官方 60k train,val 只用官方 10k test)。

用途 / Use cases

  • 目标检测 / 识别训练与微调(YOLO 格式,train.py --data yolo.yaml 直接可用)
  • 数字 OCR 方向基准与演示(见 assets/ 预览样例)
  • 与真实数据混训、域迁移实验

数据构成 / Contents

  • 尺寸:640×640 PNG;类别:0–9(10 类)
  • 标注:YOLO 归一化 txt(labels/)+ COCO JSON(annotations/,随源目录存在时)
  • 生成可复现:seed 0;溯源参数见 annotations/generation_manifest.json
  • 数据包校验和 dataset_fingerprintcbf5325c306abd761a582adededb5e031f485b068bdcae15d8369bf436bb09ed(在本 fork 仓库内复算:python3 -c "import sys;sys.path.insert(0,'data/scripts');from pathlib import Path;from make_digit_images import dataset_fingerprint;print(dataset_fingerprint(Path('<本目录>')))";跨机器只有该值相同才算同一数据包)
split 图片数 目标数 每图目标 每类目标数
train 5000 17511 2–5 0: 1744, 1: 1763, 2: 1778, 3: 1758, 4: 1773, 5: 1769, 6: 1813, 7: 1617, 8: 1734, 9: 1762
val 200 689 2–5 0: 67, 1: 69, 2: 65, 3: 78, 4: 67, 5: 74, 6: 68, 7: 63, 8: 65, 9: 73

目录结构 / Layout

YOLO_digits640/
├── README.md
├── dataset_infos.json
├── yolo.yaml          (multi 模式)
├── data/                  train.jsonl / val.jsonl(样本级索引,供平台解析与预览)
├── images/{train,val}/   *.png
├── labels/{train,val}/   *.txt
├── annotations/           COCO + generation_manifest.json
└── assets/                preview.png / sample_train_*.png

样本索引 / Sample index(data/*.jsonl

每行一个样本(JSON):image(相对本包的图片路径)、splitwidth/heightobjects[{cls, bbox}]bbox 为归一化 xywh,与 labels/*.txt 逐行同序同值)、 n_objectsprimary_linefont 字体矢量渲染 / mnist MNIST 放大着色)、backgrounddistractor_chars(该图上的形近干扰字符,按设计不标注,故不出现在 objectslabels/*.txt 里)。

快速使用 / Quick start(本仓库 fork,yaml 的 path 按项目根解析)

# 1) 取回本包(二选一),放到 yolov5 项目根下的 datasets/YOLO_digits640/
modelscope download --dataset wcpsoft/YOLO_digits640 --local_dir datasets/YOLO_digits640
hf download wcpsoft/YOLO_digits640 --repo-type=dataset --local-dir datasets/YOLO_digits640
# 2) 训练:包内 yolo.yaml 的 path 已写成 datasets/YOLO_digits640,落在上述位置即命中
python train.py --data datasets/YOLO_digits640/yolo.yaml \
    --weights yolov5n.pt --img 640 --epochs 120 --batch-size 64 --device 0

放在别处时,把 yolo.yaml 里的 path 改成该目录的绝对路径即可。

生成方式 / Reproduction

python3 data/scripts/make_digit_images.py --per-class-train 500 --per-class-val 20 --seed 0 --out datasets/mnist/generated/digits640_p500

上传 / Upload(访问令牌只走环境变量或 --token,MUST NOT 写进文件)

# HuggingFace:huggingface_hub>=1.x 的 CLI 是 hf;--repo-type 默认 model,数据集必须显式指定
hf upload wcpsoft/YOLO_digits640 <本包目录> . --repo-type=dataset --token "$HF_TOKEN"
# 魔搭:选项是 --repo-type dataset,没有 --dataset 这个写法
modelscope upload wcpsoft/YOLO_digits640 <本包目录> . --repo-type dataset

License

本数据集(图像与标注均由本项目合成生成)以 Apache License 2.0 发布。 需知悉:其中约一半数字目标由 MNIST 手写体放大着色而来,属 MNIST 的衍生内容;本包不含 MNIST 原始文件, MNIST 自身的许可条款以其来源页为准(Yann LeCun / Corinna Cortes)。

Downloads last month
148