arXiv:2501.03145v3 Announce Type: replace-cross 
Abstract: Camera-captured document images often suffer from geometric distortions caused by paper deformation, perspective distortion, and lens aberrations, significantly reducing OCR accuracy. This study develops an efficient automated method for document image dewarping that balances accuracy with computational efficiency. We propose a hybrid approach combining deep learning for document detection with classical computer vision for geometry restoration. YOLOv8 performs initial document segmentation and mask generation. Subsequently, classical CV techniques construct a topological 2D grid through cubic polynomial interpolation of document boundaries, followed by image remapping to correct nonlinear distortions. A new annotated dataset and open-source framework are provided to facilitate reproducibility and further research. Experimental evaluation against state-of-the-art methods (RectiNet, DocGeoNet, DocTr++) and mobile applications (DocScan, CamScanner, TapScanner) demonstrates superior performance. Our method achieves the lowest median Character Error Rate (CER=0.0235), Levenshtein Distance (LD=27.8), and highest Jaro--Winkler similarity (JW=0.902), approaching the quality of scanned originals. The approach requires significantly fewer computational resources and memory compared to pure deep learning solutions while delivering better OCR readability and geometry restoration quality. The proposed hybrid methodology effectively restores document geometry with computational efficiency superior to existing deep learning approaches, making it suitable for resource-constrained applications while maintaining high-quality document digitization. Project page: https://github.com/HorizonParadox/DRCCBI

تقدم هذه الدراسة طريقة آلية لتصحيح صور الوثائق الملتقطة بالكاميرا، حيث تعالج التشوهات الهندسية التي تؤثر على دقة التعرف الضوئي على الحروف (OCR). تعتمد الطريقة المقترحة على دمج التعلم العميق لاكتشاف الوثائق مع تقنيات الرؤية الحاسوبية التقليدية لاستعادة الهندسة. يتم استخدام YOLOv8 للتجزئة الأولية، تليها عملية استيفاء متعددة الحدود المكعبة لتصحيح التشوهات غير الخطية. تم تقديم مجموعة بيانات مشروحة وإطار عمل مفتوح المصدر لدعم إمكانية التكرار والبحث المستقبلي.

Este estudio presenta un método automatizado para corregir imágenes de documentos capturadas por cámara, abordando las distorsiones geométricas que afectan la precisión del OCR. La propuesta combina aprendizaje profundo para la detección de documentos y técnicas clásicas de visión por computadora para la restauración geométrica. Se utiliza YOLOv8 para la segmentación inicial, seguido de una interpolación polinómica cúbica para corregir distorsiones no lineales. Se introduce un conjunto de datos anotados y un marco de código abierto para facilitar la reproducibilidad y la investigación futura.

Cette étude présente une méthode automatisée pour déformer les images de documents capturées par caméra, en s'attaquant aux distorsions géométriques qui nuisent à la précision de l'OCR. L'approche hybride proposée utilise l'apprentissage profond pour la détection de documents et des techniques de vision par ordinateur classiques pour la restauration géométrique. YOLOv8 est utilisé pour la segmentation initiale, suivi d'une interpolation polynomiale cubique pour corriger les distorsions non linéaires. Un ensemble de données annotées et un cadre open-source sont introduits pour soutenir la repro…

This study presents an automated method for dewarping camera-captured document images, addressing geometric distortions that hinder OCR accuracy. The proposed hybrid approach utilizes deep learning for document detection and classical computer vision techniques for geometry restoration. YOLOv8 is employed for initial segmentation, followed by cubic polynomial interpolation to correct nonlinear distortions. An annotated dataset and open-source framework are introduced to support reproducibility and further research.

Efficient Document Image Dewarping via Hybrid Deep Learning and Cubic Polynomial Geometry Restoration

arXiv:2601.08401v1 Announce Type: new 
Abstract: Objectives: To overcome challenges in diagnosing pericoronitis on panoramic radiographs, an AI-assisted assessment system integrating anatomical localization, pathological classification, and interpretability. Methods: A two-stage deep learning pipeline was implemented. The first stage used YOLOv8 to detect third molars and classify their anatomical positions and angulations based on Winter's classification. Detected regions were then fed into a second-stage classifier, a modified ResNet-50 architecture, for detecting radiographic features suggestive of pericoronitis. To enhance clinical trust, Grad-CAM was used to highlight key diagnostic regions on the radiographs. Results: The YOLOv8 component achieved 92% precision and 92.5% mean average precision. The ResNet-50 classifier yielded F1-scores of 88% for normal cases and 86% for pericoronitis. Radiologists reported 84% alignment between Grad-CAM and their diagnostic impressions, supporting the radiographic relevance of the interpretability output. Conclusion: The system shows strong potential for AI-assisted panoramic assessment, with explainable AI features that support clinical confidence.

قدمت دراسة جديدة إطار عمل عميق قابل للتفسير على مرحلتين لتقييم التهاب اللثة المحيطة بالأسنان في الأشعة السينية البانورامية، باستخدام YOLOv8 لتحديد المواقع التشريحية وResNet-50 المعدل لتصنيف الأمراض. حقق النظام دقة عالية وتوافق جيد مع الانطباعات التشخيصية للأطباء، مما يعزز القابلية للتفسير من خلال تصورات Grad-CAM.

Un nuevo estudio ha presentado un marco de aprendizaje profundo explicable en dos etapas para la evaluación de la pericoronitis en radiografías panorámicas, utilizando YOLOv8 para la localización anatómica y un ResNet-50 modificado para la clasificación patológica. El sistema logró una alta precisión y una buena alineación con las impresiones diagnósticas de los radiólogos, mejorando la interpretabilidad a través de visualizaciones Grad-CAM.

Une nouvelle étude a introduit un cadre d'apprentissage profond explicable en deux étapes pour évaluer la péricoronite sur des radiographies panoramiques, utilisant YOLOv8 pour la localisation anatomique et un ResNet-50 modifié pour la classification pathologique. Le système a atteint une haute précision et une bonne concordance avec les impressions diagnostiques des radiologues, améliorant l'interprétabilité grâce aux visualisations Grad-CAM.

A new study has introduced an explainable two-stage deep learning framework for assessing pericoronitis in panoramic radiographs, utilizing YOLOv8 for anatomical localization and a modified ResNet-50 for pathological classification. The system achieved high precision and alignment with radiologists' diagnostic impressions, enhancing interpretability through Grad-CAM visualizations.

Efficient Document Image Dewarping via Hybrid Deep Learning and Cubic Polynomial Geometry Restoration

Was this article worth reading? Share it

WasItAI

DigiParser

ImgUpscaler AI

Unblurimage AI

PhotoRepair

Attentive AI

Ready to build your own newsroom?