Detection and measurement of warping in FDM additive manufacturing process using artificial intelligence and machine vision

The Fused Deposition Modeling (FDM) additive manufacturing process faces challenges such as structural defects during part fabrication. Timely detection of these defects can prevent material and time waste and, in some cases, enable process correction. This study presents an innovative approach for detecting and measuring warping defects in parts using artificial intelligence and machine vision. The proposed method allows defect detection with any type of camera (e.g., smartphone, laptop camera) under natural, non-engineered conditions in real-time. Furthermore, this approach is extendable to the detection of other visual defects.Initially, deep learning classification networks, such as VGG and Xception, were employed for defect detection. Traditional algorithms like Canny and HSV were subsequently used to measure the degree of warping. Based on the results, the process evolved toward greater reliance on intelligent methods, enabling defect detection and mask generation entirely through artificial intelligence. Ultimately, leveraging the OpenCV library and the YOLOv8 algorithm, the proposed system achieved a detection accuracy of 99% at a 0.5 threshold and an average accuracy of 0.78 in the 0.95–0.5 threshold range.To enhance the process's usability, a web-based application was developed using HTML and the Streamlit library, facilitating easier access to the system. This research represents a significant step toward integrating intelligent technologies for improving the quality of additive manufacturing processes.