license: mit
task_categories:
- feature-extraction
tags:
- ocr
- numbers
- computervision
size_categories:
- 10K<n<100K
Dataset Card for PypayaNumbers
Dataset Description
Dataset Summary
This dataset consists of images of numbers along with their bounding box coordinates and labels. The dataset is divided into train and test sets, with each set containing images, numbers, and bounding boxes. The numbers are represented as one-line text files, while the bounding boxes are in YOLO format.
Supported Tasks and Leaderboards
This dataset supports the task of Optical Character Recognition (OCR) and object detection. Specifically, it can be used for tasks like digit recognition in images.
Languages
The dataset does not contain any natural language data.
Dataset Structure
Data Instances
Each instance in the dataset comprises an image file, a corresponding text file with the number represented in the image, and a text file with the bounding box coordinates for each digit in YOLO format.
Data Fields
image: A file path to an image containing a number.number: A file path to a text file containing the number represented in the image.bounding_boxes: A file path to a text file containing the bounding box coordinates for each digit in the image.
Data Splits
The dataset is split into a training set of 5000 instances and a testing set of 2500 instances.
Dataset Creation
Curation Rationale
This dataset was curated to support the development and evaluation of models for digit recognition in images.
Source Data
Initial Data Collection and Normalization
The images and labels in this dataset were generated by taking screenshots from various computer games and programs and cutting fragments containing numbers.
Who are the source language producers?
N/A
Annotations
Annotation process
The bounding box annotations were generated using open-source LabelImg software.
Who are the annotators?
PypayaTech
Personal and Sensitive Information
The dataset does not contain any personal or sensitive information.
Considerations for Using the Data
Social Impact of Dataset
This dataset could help in improving models for digit recognition in images, which has numerous applications including automated data entry, number plate recognition, and form digitization.
Discussion of Biases
As the dataset consists of synthetic images of numbers, no inherent biases related to human demographics or behavior are expected.
Other Known Limitations
The dataset only contains images of numbers and might not generalize well to other types of characters or more complex images.
Additional Information
Dataset Curators
PypayaTech
Licensing Information
This dataset is licensed under the MIT license.
Contributions
Contributions to the dataset are welcome. Please contact the dataset curator for more information.