<![CDATA[The micro, small, and medium enterprises (MSMEs) industry plays a crucial role in supporting the Indonesian economy, yet faces challenges in production efficiency and cost management. This study presents the design and development of an IoT-based automated screen-printing system that integrates edge-based image classification using a CNN model with microcontroller-driven print actuation, specifically tailored for MSME-scale garment production. The system employs an ESP32/Raspberry Pi as the edge device, enabling local inference without cloud dependency, and utilizes MQTT protocol for IoT connectivity. Quantitative evaluation across 50 test cycles demonstrated a 96% printer success rate, a 60% reduction in production time from 45 to 18 minutes per 10 shirts, a 90% reduction in labor from 10 operators to 1, and an approximately 50% reduction in per-unit production cost from Rp65,000–80,000 to Rp30,000–40,000 per shirt. IoT connectivity testing over 48 continuous hours recorded an average MQTT latency of 120 ms and a system uptime of 98.5%, confirming the reliability of the communication layer for sustained production operations. Grounded in Industry 5.0 principles, this research advances human-machine collaboration in small-scale manufacturing contexts. The proposed system offers a cost-effective, remotely controlled, and semi-autonomous production solution, representing a novel contribution to the field of IoT-based garment manufacturing in Indonesia.]]>
Copyrights © 2026
