Deep Feature-Based Image Retrieval Using YOLOv8 and K-Nearest Neighbour Classifier

Apurva Thakur; Rajesh Kumar; Rajanish Kumar Kaushal

doi:10.1007/978-3-032-18135-0_10

Deep Feature-Based Image Retrieval Using YOLOv8 and K-Nearest Neighbour Classifier

Apurva Thakur
, Rajesh Kumar
, Rajanish Kumar Kaushal

Chandigarh University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Content-Based Image Retrieval (CBIR) is used to retrieve relevant images stored in extensive databases. It has applications in healthcare, surveillance, crime investigation, and remote sensing. This paper proposes a hybrid CBIR system that integrates deep learning-based object features using YOLOv8, texture features based on the Gray Level Co-occurrence Matrix (GLCM), and color features derived from Color Histograms into a single feature vector. The system employs the K-Nearest Neighbor (KNN) algorithm, with similarity comparison based on cosine similarity. The framework is evaluated on the Corel 1K dataset (10 categories), using precision, recall, and F1-score as performance measures. Experimental results demonstrate that the proposed approach achieves average recall, precision, and F1-score values of 80%, 77.77%, and 78.33%, respectively, outperforming several state-of-the-art CBIR systems evaluated on the same benchmark. These findings highlight the efficiency, accuracy, and computational simplicity of the system, making it suitable for large-scale, real-world image retrieval applications.

Original language	English
Title of host publication	6th Congress on Intelligent Systems - CIS 2025, Volume 5
Editors	Sandeep Kumar, Mohammad Shorif Uddin, Tarun Kumar Rawat
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	118-128
Number of pages	11
ISBN (Print)	9783032181343
DOIs	https://doi.org/10.1007/978-3-032-18135-0_10
State	Published - 2026
Externally published	Yes
Event	6th Congress on Intelligent Systems, CIS 2025 - Delhi, India Duration: 6 Sep 2025 → 7 Sep 2025

Publication series

Name	Lecture Notes in Networks and Systems
Volume	1826 LNNS
ISSN (Print)	2367-3370
ISSN (Electronic)	2367-3389

Conference

Conference	6th Congress on Intelligent Systems, CIS 2025
Country/Territory	India
City	Delhi
Period	6/09/25 → 7/09/25

Bibliographical note

Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 16 Peace, Justice and Strong Institutions

Keywords

CBIR
Color Histogram
Corel 1K
Deep Learning
F1-score
GLCM
KNN
Precision
Recall
YOLOv8

Access to Document

10.1007/978-3-032-18135-0_10

Cite this

Thakur, A., Kumar, R., & Kaushal, R. K. (2026). Deep Feature-Based Image Retrieval Using YOLOv8 and K-Nearest Neighbour Classifier. In S. Kumar, M. Shorif Uddin, & T. K. Rawat (Eds.), 6th Congress on Intelligent Systems - CIS 2025, Volume 5 (pp. 118-128). (Lecture Notes in Networks and Systems; Vol. 1826 LNNS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-032-18135-0_10

Thakur, Apurva ; Kumar, Rajesh ; Kaushal, Rajanish Kumar. / Deep Feature-Based Image Retrieval Using YOLOv8 and K-Nearest Neighbour Classifier. 6th Congress on Intelligent Systems - CIS 2025, Volume 5. editor / Sandeep Kumar ; Mohammad Shorif Uddin ; Tarun Kumar Rawat. Springer Science and Business Media Deutschland GmbH, 2026. pp. 118-128 (Lecture Notes in Networks and Systems).

@inproceedings{38b5f14bc3ae486396262c135d8bd8f7,

title = "Deep Feature-Based Image Retrieval Using YOLOv8 and K-Nearest Neighbour Classifier",

abstract = "Content-Based Image Retrieval (CBIR) is used to retrieve relevant images stored in extensive databases. It has applications in healthcare, surveillance, crime investigation, and remote sensing. This paper proposes a hybrid CBIR system that integrates deep learning-based object features using YOLOv8, texture features based on the Gray Level Co-occurrence Matrix (GLCM), and color features derived from Color Histograms into a single feature vector. The system employs the K-Nearest Neighbor (KNN) algorithm, with similarity comparison based on cosine similarity. The framework is evaluated on the Corel 1K dataset (10 categories), using precision, recall, and F1-score as performance measures. Experimental results demonstrate that the proposed approach achieves average recall, precision, and F1-score values of 80\%, 77.77\%, and 78.33\%, respectively, outperforming several state-of-the-art CBIR systems evaluated on the same benchmark. These findings highlight the efficiency, accuracy, and computational simplicity of the system, making it suitable for large-scale, real-world image retrieval applications.",

keywords = "CBIR, Color Histogram, Corel 1K, Deep Learning, F1-score, GLCM, KNN, Precision, Recall, YOLOv8",

author = "Apurva Thakur and Rajesh Kumar and Kaushal, \{Rajanish Kumar\}",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.; 6th Congress on Intelligent Systems, CIS 2025 ; Conference date: 06-09-2025 Through 07-09-2025",

year = "2026",

doi = "10.1007/978-3-032-18135-0\_10",

language = "אנגלית",

isbn = "9783032181343",

series = "Lecture Notes in Networks and Systems",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "118--128",

editor = "Sandeep Kumar and \{Shorif Uddin\}, Mohammad and Rawat, \{Tarun Kumar\}",

booktitle = "6th Congress on Intelligent Systems - CIS 2025, Volume 5",

address = "גרמניה",

}

Thakur, A, Kumar, R & Kaushal, RK 2026, Deep Feature-Based Image Retrieval Using YOLOv8 and K-Nearest Neighbour Classifier. in S Kumar, M Shorif Uddin & TK Rawat (eds), 6th Congress on Intelligent Systems - CIS 2025, Volume 5. Lecture Notes in Networks and Systems, vol. 1826 LNNS, Springer Science and Business Media Deutschland GmbH, pp. 118-128, 6th Congress on Intelligent Systems, CIS 2025, Delhi, India, 6/09/25. https://doi.org/10.1007/978-3-032-18135-0_10

Deep Feature-Based Image Retrieval Using YOLOv8 and K-Nearest Neighbour Classifier. / Thakur, Apurva; Kumar, Rajesh; Kaushal, Rajanish Kumar.
6th Congress on Intelligent Systems - CIS 2025, Volume 5. ed. / Sandeep Kumar; Mohammad Shorif Uddin; Tarun Kumar Rawat. Springer Science and Business Media Deutschland GmbH, 2026. p. 118-128 (Lecture Notes in Networks and Systems; Vol. 1826 LNNS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Deep Feature-Based Image Retrieval Using YOLOv8 and K-Nearest Neighbour Classifier

AU - Thakur, Apurva

AU - Kumar, Rajesh

AU - Kaushal, Rajanish Kumar

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

PY - 2026

Y1 - 2026

N2 - Content-Based Image Retrieval (CBIR) is used to retrieve relevant images stored in extensive databases. It has applications in healthcare, surveillance, crime investigation, and remote sensing. This paper proposes a hybrid CBIR system that integrates deep learning-based object features using YOLOv8, texture features based on the Gray Level Co-occurrence Matrix (GLCM), and color features derived from Color Histograms into a single feature vector. The system employs the K-Nearest Neighbor (KNN) algorithm, with similarity comparison based on cosine similarity. The framework is evaluated on the Corel 1K dataset (10 categories), using precision, recall, and F1-score as performance measures. Experimental results demonstrate that the proposed approach achieves average recall, precision, and F1-score values of 80%, 77.77%, and 78.33%, respectively, outperforming several state-of-the-art CBIR systems evaluated on the same benchmark. These findings highlight the efficiency, accuracy, and computational simplicity of the system, making it suitable for large-scale, real-world image retrieval applications.

AB - Content-Based Image Retrieval (CBIR) is used to retrieve relevant images stored in extensive databases. It has applications in healthcare, surveillance, crime investigation, and remote sensing. This paper proposes a hybrid CBIR system that integrates deep learning-based object features using YOLOv8, texture features based on the Gray Level Co-occurrence Matrix (GLCM), and color features derived from Color Histograms into a single feature vector. The system employs the K-Nearest Neighbor (KNN) algorithm, with similarity comparison based on cosine similarity. The framework is evaluated on the Corel 1K dataset (10 categories), using precision, recall, and F1-score as performance measures. Experimental results demonstrate that the proposed approach achieves average recall, precision, and F1-score values of 80%, 77.77%, and 78.33%, respectively, outperforming several state-of-the-art CBIR systems evaluated on the same benchmark. These findings highlight the efficiency, accuracy, and computational simplicity of the system, making it suitable for large-scale, real-world image retrieval applications.

KW - CBIR

KW - Color Histogram

KW - Corel 1K

KW - Deep Learning

KW - F1-score

KW - GLCM

KW - KNN

KW - Precision

KW - Recall

KW - YOLOv8

UR - https://www.scopus.com/pages/publications/105038985611

U2 - 10.1007/978-3-032-18135-0_10

DO - 10.1007/978-3-032-18135-0_10

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:105038985611

SN - 9783032181343

T3 - Lecture Notes in Networks and Systems

SP - 118

EP - 128

BT - 6th Congress on Intelligent Systems - CIS 2025, Volume 5

A2 - Kumar, Sandeep

A2 - Shorif Uddin, Mohammad

A2 - Rawat, Tarun Kumar

PB - Springer Science and Business Media Deutschland GmbH

T2 - 6th Congress on Intelligent Systems, CIS 2025

Y2 - 6 September 2025 through 7 September 2025

ER -

Deep Feature-Based Image Retrieval Using YOLOv8 and K-Nearest Neighbour Classifier

Abstract

Publication series

Conference

Bibliographical note

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this