The paper bag manufacturing industry confronts significant economic losses and material waste due to high rates of production defects, particularly inconsistent adhesive application. This research presents the development and validation of an intelligent defect reduction framework for the gum dotting process in multiwall paper sack production. An Internet of Things (IoT) integrated machine vision system, utilizing a YOLOv8 model trained on a custom dataset of approximately 1,300 images, was developed for real-time defect detection and classification. Through a series of simulated experiments, the system's efficacy was quantified. Spatial analysis of defect data successfully identified specific mechanical root causes, such as individual nozzle failures versus systemic pressure imbalances. A correlative analysis established a critical, non-linear relation-ship between machine velocity and defect frequency, identifying an optimal operating window of 100–160 bags/min, beyond which defect rates increase exponentially. Furthermore, the analysis of the real-time alerting protocol and subsequent maintenance interventions demonstrated the system's capacity to facilitate a strategic shift from reactive to predictive maintenance. By calculating the Mean Time Between Failures (MTBF) for specific fault-and-fix cycles, the framework provides the empirical data needed to predict component end-of-life and schedule proactive replacements. The findings confirm that this data-driven approach can significantly reduce waste, optimize production efficiency, and enhance overall product quality.