ShelfAura: A Dataset for Book Spine Detection and Recognition

Read full text Summarise with VIRAT
Authors: Godwin Wilfred Raj A, Sathish Kumar S, Pooja Shree S J, Jeswin Pio A, Sundar Ganesh C S
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This dataset, ShelfAura, is designed for the tasks of book spine detection, segmentation, and optical character recognition (OCR) in library and bookstore environments. It consists of 111 high-resolution images of bookshelves containing approximately 1,754 individual book spines. Key features: 111 raw shelf images. 1,754 manually annotated and cropped book spine images. Annotations provided for text recognition on spines. Optimized for training models like YOLO, Florence-2, Qwen and TrOCR. This dataset supports research in automated inventory management and digital library organization.

Fields of Study & Subjects

Handwritten Text Recognition Techniques Advanced Neural Network Applications Image Processing and 3D Reconstruction

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
20 March 2026
Language
EN
Document type
Dataset