AI-Enhanced Video Analysis for Edge-Based Defect Detection in Smart Manufacturing of Razor Blades

In smart manufacturing systems, integrating edge computing with AI-driven video analysis can significantly enhance real-time quality control. This paper proposes an edge-deployable framework for detecting surface defects on razor blades by combining lightweight computer vision and adaptive AI algorithms. Raw blade surface images are captured through an IoT-enabled vision system, with Otsu’s binarization and CCL-based contour analysis optimized for edge processing. To address dynamic industrial environments, we construct an adaptive background model using frame-averaging and deploy morphological filtering at the fog layer to eliminate interference. Defect classification is achieved through a collaborative edge-fog architecture, where AI-driven mask analysis determines defect types while ensuring low latency. Experimental results demonstrate that our method achieves 95.1% detection accuracy at 0.5 m/s conveyor speed with 15 ms processing latency per frame, enabling seamless integration with SDN-IoT networks for distributed quality monitoring. This approach provides a scalable solution for real-time defect detection in future-ready smart factories.