Journal of Confy Demonstration Studies

Deep Learning for Rice Leaf Disease Classification in Smallholder Farms

This study proposes a lightweight convolutional neural network for classifying rice leaf diseases from smartphone images captured in Indonesian smallholder farms. Trained on 8,400 field images across four disease classes, the model attains 96.2% accuracy while running on-device, enabling offline diagnosis where connectivity is limited.

A Federated Query Layer for Multi-Tenant Academic Conference Systems

Multi-tenant conference platforms must enforce strict data isolation while sharing infrastructure. We present a row-level federated query layer that injects tenant scope at the ORM boundary, eliminating a class of cross-tenant leakage bugs. Benchmarks on a 40-tenant deployment show under 6% overhead versus hand-written scoped queries.

Open-Access Journal Hosting for Indonesian Campuses: A Managed-Service Model

Accreditation readiness (SINTA, DOAJ, Garuda) remains a barrier for small Indonesian journals lacking technical staff. We describe a managed OJS hosting model that bundles deployment, updates, backups, and metadata compliance, and report on onboarding time and indexing outcomes across pilot journals.