|
|
--- |
|
|
license: apache-2.0 |
|
|
task_categories: |
|
|
- image-to-text |
|
|
language: |
|
|
- en |
|
|
tags: |
|
|
- image |
|
|
- pdf |
|
|
- document |
|
|
- receipt |
|
|
size_categories: |
|
|
- n<1K |
|
|
--- |
|
|
|
|
|
# Openpdf-Blank-v2.0 |
|
|
|
|
|
Openpdf-Blank-v2.0 is a small dataset containing blank or near-blank PDF image samples. This dataset is primarily designed to help train and evaluate document processing models, especially in tasks like: |
|
|
|
|
|
* Identifying and filtering blank or noise-filled documents. |
|
|
* Preprocessing stages for OCR pipelines. |
|
|
* Receipt/document classification tasks. |
|
|
|
|
|
## Dataset Structure |
|
|
|
|
|
* **Modality**: Image |
|
|
* **Languages**: English (if applicable) |
|
|
* **Size**: Less than 1,000 samples |
|
|
* **License**: Apache-2.0 |
|
|
|
|
|
## Usage |
|
|
|
|
|
```python |
|
|
from datasets import load_dataset |
|
|
|
|
|
dataset = load_dataset("prithivMLmods/Openpdf-Blank-v2.0") |
|
|
``` |
|
|
|
|
|
## Intended Use |
|
|
|
|
|
The dataset is ideal for training or benchmarking models that need to recognize and filter out blank pages in document images, such as in: |
|
|
|
|
|
* Invoice preprocessing pipelines |
|
|
* Bulk document OCR systems |
|
|
* Smart scanning tools |
