Receipt Images with OCR and Key Information Extraction Labels

CV is a collection of receipt images for benchmarking key information extraction from documents. The dataset provides OCR outputs including bounding boxes, recognized text, and class labels for each receipt. It was created by nvm472001 and last updated in December 2022.

Use Cases

• Train a model to extract key information from receipt images using the provided OCR text and bounding box coordinates.
• Benchmark key information extraction performance by predicting the class labels associated with recognized text segments.
• Develop a layout-aware document understanding model using the combination of image data, OCR text, and bounding box information.

Strengths

• Dataset is associated with a published research paper (arXiv:2103.14470), providing academic context.
• Last update recorded was December 13, 2022, indicating a specific snapshot.

Limitations

• The dataset size, number of rows, and specific column structure are unknown.
• The geographic origin of the receipts and the time range of data collection are unspecified.

Provenance

Source

huggingface

Collection Method

Collection of receipt photos with associated OCR outputs and class labels for key information extraction.

Freshness

Last updated on 2022-12-13.