Research

Research &
Publications.

Peer-reviewed publication in computer vision and educational systems. Active research interests across AI in EdTech, adaptive learning analytics, IS adoption in emerging economies, and responsible governance of educational data — informed by three years of nationwide LMS and AI workflow deployments.

Publications

Peer-Reviewed2023·Computer Vision · EdTech

An Automated Attendance System Using Facial Detection and Recognition Technology

Jha, P. B., Basnet, A., Pokhrel, B., Pokhrel, B., Thakur, G. K., & Chhetri, S.
Apex Journal of Business and Management, Vol. 1, No. 1, 2023, pp. 103–120.

Abstract

This paper presents the design and implementation of a real-time automated attendance system for educational institutions using facial detection and recognition technology. The system integrates Haar Cascade for face detection and the Local Binary Patterns Histogram (LBPH) algorithm for face recognition, enabling accurate student identification and attendance marking without manual intervention. The study addresses key challenges in deploying computer vision systems in real classroom environments — including lighting variability, occlusion, and teacher adoption — and proposes a practical architecture balancing computational efficiency with recognition accuracy.

Algorithms

Haar Cascade · LBPH

Domain

Computer Vision · Education

Journal

Apex Journal of Business and Management

Google Scholar ResearchGate

Research Interests

01

Artificial Intelligence in Educational Technology

How AI systems — from adaptive assessment engines to learning analytics dashboards — can be responsibly designed and deployed in classroom environments, particularly in resource-constrained settings.

02

Adaptive Learning Analytics & Standards-Driven LMS

The design of learning management systems that respond to individual learner behaviour and produce analytics that teachers and administrators can act on — built to shared data standards rather than vendor-specific silos.

03

Information Systems Adoption in Emerging Economies

Structural and behavioural factors that predict whether an EdTech or enterprise system moves beyond its pilot deployment — drawing on IS diffusion theory and empirical evidence from South Asian contexts.

04

AI-Driven Business Process Automation

The design of LLM-based workflow orchestration systems for high-value, structured business processes — with emphasis on output reliability, human-in-the-loop design, and measurable efficiency outcomes.

05

Human-Computer Interaction for Low-Resource Contexts

Interface and interaction design principles for users in environments with limited bandwidth, intermittent connectivity, lower digital literacy, and affordable mid-range hardware.

06

Technology Entrepreneurship & Digital Transformation

How technology companies in developing economies can build durable, export-capable digital products — and how educational institutions can use digital transformation to improve student and teacher outcomes.

Open Questions

The questions I'm working on.

These are the open questions that shape my work in EdTech and AI for emerging economies. They came out of three years of LMS deployments, AI workflow design, and digital transformation projects in Nepal.

01

How can personalised, adaptive learning be implemented in environments where bandwidth, devices, and teacher training are all limited?

Designing for constraint rather than optimum — the question that shaped every LMS deployment in Nepal.

02

What structural factors predict whether an LMS deployment in an emerging economy matures beyond its pilot?

Five deployments, five different definitions of a gradebook. The bottleneck is not the technology.

03

How do we govern educational data responsibly so that AI-driven insight does not quietly amplify the inequities it was meant to close?

As AI enters Nepali classrooms, the governance frameworks need to arrive first — not after.

04

What does production-grade LLM workflow design look like outside Silicon Valley — when the budget, team, and infrastructure are all constrained?

n8n + Claude is the practical answer for now. The harder question is what becomes the durable abstraction.

Academic Profiles

Find publications, citations, and research profile on:

Google Scholar ResearchGate LinkedIn