Design and Evaluation of a REST API Backend with Payment Integration for AR Self-Service Retail

Abstract

Augmented reality (AR) virtual try-on has become common in self-service retail, yet most research still concentrates on the frontend, and the server side and its measured performance receive little attention. This study designed, implemented, and evaluated a REST API backend for AR retail transactions, including end-to-end payment, in a smart-mirror deployment. Built on Laravel, the backend exposes a single JSON contract to both the web-based AR try-on client and an administrative dashboard, protects every privileged endpoint with JWT, and integrates a Midtrans Snap gateway through an asynchronous, SHA-512-verified webhook that reconciles each transaction server-to-server, independently of client connectivity. The backend was evaluated using four methods: black-box testing of twelve CRUD modules, five formal integration scenarios, a load test from one to one hundred concurrent users (about four thousand requests), and a structured expert review. Under the intended single-station workload, login averaged about 0.4 s at a 0.83% error rate and catalog reads about 0.15 s with no errors; the read endpoints remained error-free up to fifty concurrent users, whereas the password-hashing path saturated first. Three experts rated the system 120 of 135 (“very good”). These results apply to the single-station scenario tested, not to multi-store or high-concurrency deployment.