---
language:
- en
license: other
size_categories:
- 1K<n<10K
task_categories:
- image-classification
- object-detection
paperswithcode_id: rvl-cdip
pretty_name: RVL-CDIP Mini
dataset_info:
  features:
  - name: image
    dtype: image
  - name: width
    dtype: int64
  - name: height
    dtype: int64
  - name: category
    dtype: string
  - name: ocr_words
    sequence: string
  - name: word_boxes
    sequence:
      sequence: int64
  - name: ocr_paragraphs
    sequence: string
  - name: paragraph_boxes
    sequence:
      sequence: int64
  - name: label
    dtype: int64
  splits:
  - name: train
    num_bytes: 353331082.8
    num_examples: 3200
  - name: validation
    num_bytes: 43966539.0
    num_examples: 400
  - name: test
    num_bytes: 42947141.0
    num_examples: 400
  download_size: 424911398
  dataset_size: 440244762.8
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
---

# Dataset Card for RVL-CDIP-MINI

**This dataset is a subset (1%) of the original [aharley/rvl_cdip](https://huggingface.co/datasets/aharley/rvl_cdip) merged with the corresponding annotations from [jordyvl/rvl_cdip_easyocr](https://huggingface.co/datasets/jordyvl/rvl_cdip_easyocr).**

You can easily and quickly load it:

```python
dataset = load_dataset("dvgodoy/rvl_cdip_mini")
```

```
DatasetDict({
    train: Dataset({
        features: ['image', 'width', 'height', 'category', 'ocr_words', 'word_boxes', 'ocr_paragraphs', 'paragraph_boxes', 'label'],
        num_rows: 3200
    })
    validation: Dataset({
        features: ['image', 'width', 'height', 'category', 'ocr_words', 'word_boxes', 'ocr_paragraphs', 'paragraph_boxes', 'label'],
        num_rows: 400
    })
    test: Dataset({
        features: ['image', 'width', 'height', 'category', 'ocr_words', 'word_boxes', 'ocr_paragraphs', 'paragraph_boxes', 'label'],
        num_rows: 400
    })
})
```

## Table of Contents
- [Dataset Description](#dataset-description)
  - [Dataset Summary](#dataset-summary)
  - [Supported Tasks](#supported-tasks-and-leaderboards)
  - [Languages](#languages)
- [Dataset Structure](#dataset-structure)
  - [Data Instances](#data-instances)
  - [Data Fields](#data-instances)
  - [Data Splits](#data-instances)
- [Dataset Creation](#dataset-creation)
  - [Curation Rationale](#curation-rationale)
  - [Source Data](#source-data)
  - [Annotations](#annotations)
  - [Personal and Sensitive Information](#personal-and-sensitive-information)
- [Considerations for Using the Data](#considerations-for-using-the-data)
  - [Social Impact of Dataset](#social-impact-of-dataset)
  - [Discussion of Biases](#discussion-of-biases)
  - [Other Known Limitations](#other-known-limitations)
- [Additional Information](#additional-information)
  - [Dataset Curators](#dataset-curators)
  - [Licensing Information](#licensing-information)
  - [Citation Information](#citation-information)

## Dataset Description

- **Homepage:** [The RVL-CDIP Dataset](https://www.cs.cmu.edu/~aharley/rvl-cdip/)
- **Repository:**
- **Paper:** [Evaluation of Deep Convolutional Nets for Document Image Classification and Retrieval](https://arxiv.org/abs/1502.07058)
- **Leaderboard:** [RVL-CDIP leaderboard](https://paperswithcode.com/dataset/rvl-cdip)
- **Point of Contact:** [Adam W. Harley](mailto:aharley@cmu.edu)

### Dataset Summary

The original RVL-CDIP (Ryerson Vision Lab Complex Document Information Processing) dataset consists of 400,000 grayscale images in 16 classes, with 25,000 images per class. There are 320,000 training images, 40,000 validation images, and 40,000 test images. The images are sized so their largest dimension does not exceed 1000 pixels.

**This "mini" version contains only the first 4,000 images from the original dataset: 3,200 training images, 400 validation images, and 400 test images.**

### Supported Tasks and Leaderboards

- `image-classification`: The goal of this task is to classify a given document into one of 16 classes representing document types (letter, form, etc.). The leaderboard for this task is available [here](https://paperswithcode.com/sota/document-image-classification-on-rvl-cdip).

### Languages

All the classes and documents use English as their primary language.

## Dataset Structure

### Data Instances

A sample from the training set is provided below :
```
{
    'image': <PIL.TiffImagePlugin.TiffImageFile image mode=L size=754x1000 at 0x7F9A5E92CA90>,
    'width': 754,
    'height': 1000,
    'category': 'advertisement',
    'ocr_words': [...],
    'word_boxes': [[...]],
    'ocr_paragraphs': [...],
    'paragraph_boxes': [[...]],
    'label': 4
}
```

### Data Fields

- `image`: A `PIL.Image.Image` object containing a document.
- `width`: image's width.
- `height`: image's height.
- `category`: class label.
- `ocr_words`: list of OCRed words.
- `word_boxes`: list of box coordinates in `(xmin, ymin, xmax, ymax)` format (Pascal VOC).
- `ocr_paragraphs`: list of OCRed paragraphs.
- `paragraph_boxes`: list of box coordinates in `(xmin, ymin, xmax, ymax)` format (Pascal VOC).
- `label`: an `int` classification label.

<details>
  <summary>Class Label Mappings</summary>

```json
{
  "0": "letter",
  "1": "form",
  "2": "email",
  "3": "handwritten",
  "4": "advertisement",
  "5": "scientific report",
  "6": "scientific publication",
  "7": "specification",
  "8": "file folder",
  "9": "news article",
  "10": "budget",
  "11": "invoice",
  "12": "presentation",
  "13": "questionnaire",
  "14": "resume",
  "15": "memo"
}
```

</details>

### Data Splits

|   |train|test|validation|
|----------|----:|----:|---------:|
|# of examples|3200|400|400|

The dataset was split in proportions similar to those of ImageNet.
- 3200 images were used for training,
- 400 images for validation, and 
- 400 images for testing. 

## Dataset Creation

### Curation Rationale

From the paper:
> This work makes available a new labelled subset of the IIT-CDIP collection, containing 400,000
document images across 16 categories, useful for training new CNNs for document analysis.

### Source Data

#### Initial Data Collection and Normalization

The same as in the IIT-CDIP collection.

#### Who are the source language producers?

The same as in the IIT-CDIP collection.

### Annotations

#### Annotation process

The same as in the IIT-CDIP collection.

#### Who are the annotators?

The same as in the IIT-CDIP collection.

### Personal and Sensitive Information

[More Information Needed]

## Considerations for Using the Data

### Social Impact of Dataset

[More Information Needed]

### Discussion of Biases

[More Information Needed]

### Other Known Limitations

[More Information Needed]

## Additional Information

### Dataset Curators

The dataset was curated by the authors - Adam W. Harley, Alex Ufkes, and Konstantinos G. Derpanis.

### Licensing Information

RVL-CDIP is a subset of IIT-CDIP, which came from the [Legacy Tobacco Document Library](https://www.industrydocuments.ucsf.edu/tobacco/), for which license information can be found [here](https://www.industrydocuments.ucsf.edu/help/copyright/).

### Citation Information

```bibtex
@inproceedings{harley2015icdar,
    title = {Evaluation of Deep Convolutional Nets for Document Image Classification and Retrieval},
    author = {Adam W Harley and Alex Ufkes and Konstantinos G Derpanis},
    booktitle = {International Conference on Document Analysis and Recognition ({ICDAR})}},
    year = {2015}
}
```

### Contributions

Thanks to [@dnaveenr](https://github.com/dnaveenr) for adding this dataset.