An Accurate Book Spine Detection Network Based on Improved Oriented R-CNN

Book localization is crucial for the development of intelligent book inventory systems, where the high-precision detection of book spines is a critical requirement. However, the varying tilt angles and diverse aspect ratios of books on library shelves often reduce the effectiveness of conventional o...

Full description

Saved in:
Bibliographic Details
Main Authors: Haibo Ma, Chaobo Wang, Ang Li, Aide Xu, Dong Han
Format: Article
Language:English
Published: MDPI AG 2024-12-01
Series:Sensors
Subjects:
Online Access:https://www.mdpi.com/1424-8220/24/24/7996
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850037279896109056
author Haibo Ma
Chaobo Wang
Ang Li
Aide Xu
Dong Han
author_facet Haibo Ma
Chaobo Wang
Ang Li
Aide Xu
Dong Han
author_sort Haibo Ma
collection DOAJ
description Book localization is crucial for the development of intelligent book inventory systems, where the high-precision detection of book spines is a critical requirement. However, the varying tilt angles and diverse aspect ratios of books on library shelves often reduce the effectiveness of conventional object detection algorithms. To address these challenges, this study proposes an enhanced oriented R-CNN algorithm for book spine detection. First, we replace the standard 3 × 3 convolutions in ResNet50’s residual blocks with deformable convolutions to enhance the network’s capacity for modeling the geometric deformations of book spines. Additionally, the PAFPN (Path Aggregation Feature Pyramid Network) was integrated into the neck structure to enhance multi-scale feature fusion. To further optimize the anchor box design, we introduce an adaptive initial cluster center selection method for K-median clustering. This allows for a more accurate computation of anchor box aspect ratios that are better aligned with the book spine dataset, enhancing the model’s training performance. We conducted comparison experiments between the proposed model and other state-of-the-art models on the book spine dataset, and the results demonstrate that the proposed approach reaches an mAP of 90.22%, which outperforms the baseline algorithm by 4.47 percentage points. Our method significantly improves detection accuracy, making it highly effective for identifying book spines in real-world library environments.
format Article
id doaj-art-649aedf3a9824908b8bd223d0d3c9f37
institution DOAJ
issn 1424-8220
language English
publishDate 2024-12-01
publisher MDPI AG
record_format Article
series Sensors
spelling doaj-art-649aedf3a9824908b8bd223d0d3c9f372025-08-20T02:56:55ZengMDPI AGSensors1424-82202024-12-012424799610.3390/s24247996An Accurate Book Spine Detection Network Based on Improved Oriented R-CNNHaibo Ma0Chaobo Wang1Ang Li2Aide Xu3Dong Han4Library, Panjin Campus of Dalian University of Technology, Panjin 124000, ChinaSchool of Information Science and Technology, Dalian Maritime University, Dalian 116026, ChinaSchool of Information Science and Technology, Dalian Maritime University, Dalian 116026, ChinaSchool of Information Science and Technology, Dalian Maritime University, Dalian 116026, ChinaLibrary, Panjin Campus of Dalian University of Technology, Panjin 124000, ChinaBook localization is crucial for the development of intelligent book inventory systems, where the high-precision detection of book spines is a critical requirement. However, the varying tilt angles and diverse aspect ratios of books on library shelves often reduce the effectiveness of conventional object detection algorithms. To address these challenges, this study proposes an enhanced oriented R-CNN algorithm for book spine detection. First, we replace the standard 3 × 3 convolutions in ResNet50’s residual blocks with deformable convolutions to enhance the network’s capacity for modeling the geometric deformations of book spines. Additionally, the PAFPN (Path Aggregation Feature Pyramid Network) was integrated into the neck structure to enhance multi-scale feature fusion. To further optimize the anchor box design, we introduce an adaptive initial cluster center selection method for K-median clustering. This allows for a more accurate computation of anchor box aspect ratios that are better aligned with the book spine dataset, enhancing the model’s training performance. We conducted comparison experiments between the proposed model and other state-of-the-art models on the book spine dataset, and the results demonstrate that the proposed approach reaches an mAP of 90.22%, which outperforms the baseline algorithm by 4.47 percentage points. Our method significantly improves detection accuracy, making it highly effective for identifying book spines in real-world library environments.https://www.mdpi.com/1424-8220/24/24/7996book spine detectionoriented R-CNNdeformable convolutionsecondary feature fusionK-median clustering
spellingShingle Haibo Ma
Chaobo Wang
Ang Li
Aide Xu
Dong Han
An Accurate Book Spine Detection Network Based on Improved Oriented R-CNN
Sensors
book spine detection
oriented R-CNN
deformable convolution
secondary feature fusion
K-median clustering
title An Accurate Book Spine Detection Network Based on Improved Oriented R-CNN
title_full An Accurate Book Spine Detection Network Based on Improved Oriented R-CNN
title_fullStr An Accurate Book Spine Detection Network Based on Improved Oriented R-CNN
title_full_unstemmed An Accurate Book Spine Detection Network Based on Improved Oriented R-CNN
title_short An Accurate Book Spine Detection Network Based on Improved Oriented R-CNN
title_sort accurate book spine detection network based on improved oriented r cnn
topic book spine detection
oriented R-CNN
deformable convolution
secondary feature fusion
K-median clustering
url https://www.mdpi.com/1424-8220/24/24/7996
work_keys_str_mv AT haiboma anaccuratebookspinedetectionnetworkbasedonimprovedorientedrcnn
AT chaobowang anaccuratebookspinedetectionnetworkbasedonimprovedorientedrcnn
AT angli anaccuratebookspinedetectionnetworkbasedonimprovedorientedrcnn
AT aidexu anaccuratebookspinedetectionnetworkbasedonimprovedorientedrcnn
AT donghan anaccuratebookspinedetectionnetworkbasedonimprovedorientedrcnn
AT haiboma accuratebookspinedetectionnetworkbasedonimprovedorientedrcnn
AT chaobowang accuratebookspinedetectionnetworkbasedonimprovedorientedrcnn
AT angli accuratebookspinedetectionnetworkbasedonimprovedorientedrcnn
AT aidexu accuratebookspinedetectionnetworkbasedonimprovedorientedrcnn
AT donghan accuratebookspinedetectionnetworkbasedonimprovedorientedrcnn