Research Article
A Graph-Enhanced Multimodal Transformer Model for
Fine-Grained Document Parsing
Gabriel Ihuoma Lilian*,
Laud Charles Ochei
,
Martha Ozohu Musa
Issue:
Volume 15, Issue 1, February 2026
Pages:
1-18
Received:
31 July 2026
Accepted:
17 August 2026
Published:
9 September 2026
Abstract: The digitization of documents across industries has created an urgent need for intelligent systems capable of extracting structured information from unstructured layouts. While transformer-based models like LayoutLMv3 have advanced document understanding, they struggle to capture relational dependencies between spatially separated but semantically related elements—a fundamental challenge in form understanding and layout analysis. This paper introduces IntelliDocFormer, an enhanced multimodal transformer that explicitly models relationships between document elements through graph-based relational reasoning. Building on LayoutLMv3, our architecture incorporates a Graph Attention Network encoder that captures both sequential and spatial dependencies between text tokens, complemented by document-type classification for adaptive processing. We evaluate IntelliDocFormer on DocBank (4,000 samples) and FUNSD (119 samples), achieving Macro-F1 scores of 0.7435 (DocBank) and 0.7450 (FUNSD); performance improved substantially when scaling from 50 to 4,000 DocBank training samples (a 6,045% relative gain over a near-zero low-data baseline). Ablation studies confirm that graph-based reasoning contributes the most significant performance gain (+0.04 Macro-F1). Our results demonstrate that explicit relationship modeling is critical for fine-grained document parsing, particularly for structurally complex elements like tables (F1=0.94) and paragraphs (F1=0.75). On the validated element types (tables, paragraphs), inference runs at 1.67 sec/sample on a single Tesla T4 GPU. These findings establish graph-enhanced multimodal transformers as a promising direction for document intelligence, with implications for automated data extraction across finance, healthcare, and legal domains, particularly for well-represented structural elements such as tables and paragraphs; performance on low-frequency structural classes (e.g., footers, section headers) currently limits applicability to compliance-sensitive extraction tasks. To probe generalization beyond these two domains, we additionally repurposed IntelliDocFormer's document-type classification head for 16-way real-world document classification on RVL-CDIP using a stratified 3,200-image subsample. This yielded a Macro-F1 of 0.22 (Accuracy 0.29) — well below both our DocBank/FUNSD results and literature-reported RVL-CDIP figures for other architectures — revealing that the document-type module, as configured here, does not yet generalize to fine-grained, small-sample, multi-class real-world classification; we report this result in full, together with its diagnostic analysis, as an explicit boundary condition on our generalization claims.
Abstract: The digitization of documents across industries has created an urgent need for intelligent systems capable of extracting structured information from unstructured layouts. While transformer-based models like LayoutLMv3 have advanced document understanding, they struggle to capture relational dependencies between spatially separated but semantically ...
Show More
Research Article
Fine-Tuned YOLOv8n for Automated Helmet and Seatbelt Compliance Detection in Resource-Constrained Road Safety Enforcement
Nnanna Ekedebe*
Issue:
Volume 15, Issue 1, February 2026
Pages:
19-25
Received:
13 July 2026
Accepted:
24 July 2026
Published:
18 September 2026
Abstract: Road traffic fatalities remain a significant global public health crisis, with the WHO identifying helmet and seatbelt non-compliance as among the leading contributing risk factors. According to the WHO Global Status Report on Road Safety 2023, Nigeria is part of WHO African region, which has the among the highest road traffic death rates in the world at 19.4 deaths per 100,000 people, with enforcement of helmet and seatbelt compliance remaining persistently limited by personnel constraints and the scale of the road network. With over 195,000 kilometers of road network to monitor, manual enforcement by Federal Road Safety Corps officers alone is not scalable, motivating the need for an automated, resource-efficient screening approach. This study fine-tunes a pre-trained YOLOv8n object detection model for helmet and seatbelt compliance detection using the publicly available SeatbeltHelmet dataset comprising 9,166 images across 9 annotated classes. The dataset was split into 8,020 training, 764 validation and 382 test images, and the model was trained for 50 epochs with a batch size of 16 and an image size of 640×640 pixels using the AdamW optimizer on a free-tier Google Colab GPU. The model was evaluated on a held-out test set of 382 images using mAP50, mAP50-95, Precision and Recall. The fine-tuned model achieved an overall mAP50 of 0.599, with Person-Seatbelt achieving the strongest per-class performance at mAP50 of 0.912 and Person-NoSeatbelt at 0.847. Helmet and No-Helmet detection performed comparatively weaker, at mAP50 of 0.537 and 0.380 respectively, reflecting the smaller object size and visual similarity of helmets at typical traffic camera distances. These results demonstrate that a lightweight fine-tuned object detection model is viable for automated road safety compliance screening in resource-constrained African settings where manual monitoring alone is insufficient to address the scale of the problem.
Abstract: Road traffic fatalities remain a significant global public health crisis, with the WHO identifying helmet and seatbelt non-compliance as among the leading contributing risk factors. According to the WHO Global Status Report on Road Safety 2023, Nigeria is part of WHO African region, which has the among the highest road traffic death rates in the wo...
Show More
