Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM66337
Volume:
10 |
7 |
Jul 2026
Pages:
1-5
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.
Department of Computer Science and Applications
JSPM University, Pune
Department of Computer Science and Applications,
JSPM University, Pune, India
Motor vehicle insurance in India continues to rely on claim-handling procedures that were designed for a much smaller and slower-moving market: an adjuster inspects the vehicle, paperwork is gathered and checked by hand, and fraud is screened using rules of thumb built from experience rather than data. Individually, machine learning, computer vision, and natural language processing have each been shown capable of speeding up one piece of this workflow damage estimation, document reading, or fraud scoring yet the published work in this space almost always treats these three pieces as separate problems, leaving a person to stitch the outputs together and make the final call. In this paper, we present a preliminary, concept-stage contribution that precedes our full technical study. We examine the reasons claims automation research has remained fragmented, set out the requirements that a genuinely end-to-end system would need to satisfy, and lay out a conceptual architecture spanning intake, vision-based damage assessment, document understanding, fraud scoring, and a unifying decision layer that we intend to close this gap with. Rather than reporting experimental results, we establish the problem formulation, position our proposal against existing literature, and outline the methodology, datasets, and evaluation plan we will use in our subsequent full-length study, in which we build and evaluate the system.
KEYWORDS
Manual vehicle-insurance claims processing in India remains slow, inconsistent, and vulnerable to fraud that is difficult for individual reviewers to catch. While machine learning, computer vision and natural language processing have each been shown to help with one piece of this workflow, the literature has not yet converged on a design in which these components are integrated into a single automated decision-making system. In this paper, we made the case for such a system: we reviewed the closest prior work to establish where it falls short of full integration, and we proposed a five-stage conceptual architecture intake, vision-based damage assessment, document understanding, fraud scoring, and an automated decision layer intended to close that gap. As the next stage of this research, we will implement and empirically evaluate this architecture in a full-length a design in which these components are integrated and measure the modules described here against the manual baseline they are intended to improve upon.
[1] Ministry of Road Transport and Highways (MoRTH), Road Transport Year Book 2022–2023, Government of India. [2] Insurance Regulatory and Development Authority of India (IRDAI), Annual Report 2023–2024, Hyderabad, India. [3] Insurance Information Institute (III), Auto Insurance Claims Process, 2023. [4] McKinsey & Company, Insurance 2030 — The Impact of AI on Claims Management, 2021. [5] K. Rahul, "Automated Vehicle and Property Loss Assessment Using Convolutional Neural Networks," IEEE Trans. Comput. Intell. AI Games, vol. 12, no. 3, pp. 88–99, 2020. [6] M. Shaik, "Intelligent Automation in Insurance: Machine Learning Models for Fraud Detection and Predictive Analytics," J. Insurance Technol., vol. 7, no. 2, pp. 45–62, 2019. [7] Coalition Against Insurance Fraud, The Cost of Insurance Fraud, 2023. [8] J. Redmon and A. Farhadi, "YOLOv3: An Incremental Improvement," arXiv:1804.02767, 2018. [9] K. He, X. Zhang, S. Ren, and J. Sun, "Deep Residual Learning for Image Recognition," in Proc. IEEE CVPR, 2016, pp. 770–778. [10] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding," in Proc. NAACL-HLT, 2019, pp. 4171–4186.
Publication Date
7/25/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
0
Total views since publication
PDF Downloads
7
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...