Openpdf-Blank-v2.0-Sample

# Openpdf-Blank-v2.0-Sample

**Openpdf-Blank-v2.0-Sample** is a sample dataset of blank or near-blank invoice and receipt documents. It contains 255 high-resolution scanned images extracted and cleaned from document PDFs. This dataset is intended to support training and evaluation of OCR, document classification, and layout-based filtering models where blank or structurally minimal pages must be identified and processed.

## Dataset Summary

* **Format**: Parquet (auto-converted)
* **Modality**: Image
* **Size**: 84.8 MB
* **Number of Samples**: 255
* **Split**:

* `train`: 255 images
* **Image Dimensions**: Approximately 1690 x 1690 px
* **License**: Apache 2.0

## Features

* Contains scanned images of documents with minimal content or structural layout only.
* Suitable for:

* Blank page detection
* Document filtering
* Pre-processing pipeline validation
* Background noise training for OCR tasks

## How to Use

You can load the dataset using the Hugging Face `datasets` library:

```python
from datasets import load_dataset

dataset = load_dataset("prithivMLmods/Openpdf-Blank-v2.0-Sample")

# Access the first image
image = dataset["train"][0]["image"]
image.show()
```

Each record in the dataset contains:

* `image`: A PIL.Image object of the scanned blank/near-blank page.

## Use Cases

* Training models to detect and discard blank or non-informative pages in document workflows.
* Evaluating the robustness of OCR pipelines to blank document noise.
* Dataset balancing for invoice or receipt classifiers.

# Openpdf-Blank-v2.0-Sample

**Openpdf-Blank-v2.0-Sample** 是空白或近空白发票与收据文档的示例数据集，包含从PDF文档中提取并清洗后的255张高分辨率扫描图像。本数据集旨在支持光学字符识别（Optical Character Recognition，OCR）、文档分类以及基于布局的过滤模型的训练与评估，这类模型需识别并处理空白或结构极简的页面。

## 数据集概览
* **格式**：Parquet（自动转换格式）
* **模态**：图像
* **大小**：84.8 MB
* **样本数量**：255
* **划分**：
* `train`：255张图像
* **图像尺寸**：约1690 × 1690像素
* **许可证**：Apache 2.0

## 数据集特性
* 包含仅含少量内容或极简结构布局的文档扫描图像。
* 适用场景：
* 空白页面检测
* 文档过滤
* 预处理流水线验证
* OCR任务的背景噪声训练

## 使用方法
可通过Hugging Face `datasets`库加载该数据集：
python
from datasets import load_dataset

dataset = load_dataset("prithivMLmods/Openpdf-Blank-v2.0-Sample")

# 访问第一张图像
image = dataset["train"][0]["image"]
image.show()

数据集中每条记录包含：
* `image`：扫描得到的空白/近空白页面的PIL.Image对象。

## 应用场景
* 训练模型以在文档工作流中检测并丢弃空白或无信息价值的页面。
* 评估OCR流水线对空白文档噪声的鲁棒性。
* 用于发票或收据分类器的数据集均衡。