Comparative Performance of YOLOv8, YOLOv9, YOLOv10, and YOLOv11 for Layout Analysis of Historical Documents Images DOI Creative Commons

Eder Silva dos Santos Júnior,

Thuanne Paixão, Ana Beatriz Alvarez

и другие.

Applied Sciences, Год журнала: 2025, Номер 15(6), С. 3164 - 3164

Опубликована: Март 14, 2025

The digitalization of historical documents is interest for many reasons, including preservation, accessibility, and searchability. One the main challenges with digitization old newspapers involves complex layout analysis, where content types document must be determined. In this context, paper presents an evaluation most recent YOLO methods analysis layouts. Initially, a new dataset called BHN was created made available, standing out as first Brazilian detection. experiments were held using YOLOv8, YOLOv9, YOLOv10, YOLOv11 architectures. For training, validation, testing models, following newspaper datasets combined: BHN, GBN, Printed BlaLet GT. Recall, precision, mean average precision (mAP) used to evaluate performance models. results indicate that best performer Recalltest 81% mAPtest 89%. This provides insights on advantages these models in detection also promotes improvement image conversion into editable accessible formats.

Язык: Английский

Comparative Result Analysis of Cauliflower Disease Classification Based on Deep Learning Approach VGG16, Inception v3, ResNet, and a Custom CNN Model DOI Creative Commons

Asif Shahriar Arnob,

Ashfakul Karim Kausik,

Zohirul Islam

и другие.

Hybrid Advances, Год журнала: 2025, Номер unknown, С. 100440 - 100440

Опубликована: Март 1, 2025

Язык: Английский

Процитировано

0
Comparative Performance of YOLOv8, YOLOv9, YOLOv10, and YOLOv11 for Layout Analysis of Historical Documents Images DOI Creative Commons

Eder Silva dos Santos Júnior,

Thuanne Paixão, Ana Beatriz Alvarez

и другие.

Applied Sciences, Год журнала: 2025, Номер 15(6), С. 3164 - 3164

Опубликована: Март 14, 2025

The digitalization of historical documents is interest for many reasons, including preservation, accessibility, and searchability. One the main challenges with digitization old newspapers involves complex layout analysis, where content types document must be determined. In this context, paper presents an evaluation most recent YOLO methods analysis layouts. Initially, a new dataset called BHN was created made available, standing out as first Brazilian detection. experiments were held using YOLOv8, YOLOv9, YOLOv10, YOLOv11 architectures. For training, validation, testing models, following newspaper datasets combined: BHN, GBN, Printed BlaLet GT. Recall, precision, mean average precision (mAP) used to evaluate performance models. results indicate that best performer Recalltest 81% mAPtest 89%. This provides insights on advantages these models in detection also promotes improvement image conversion into editable accessible formats.

Язык: Английский

Процитировано

0