Counting Method of Instant Noodle Products Outer Packaging Based on Improved YOLOV8

This paper thoroughly investigates the application of an optimized YOLOV8 deep learning framework in high-precision recognition and counting of instant rice noodle (luosifen) packaging. Considering the unique characteristics of luosifen packaging, we adopted a SOTA (State-of-the-Art) object detection network spanning resolutions from P5 640 to P6 1280, incorporating the. By introducing multi-scale model designs such as, we achieved flexible deployment from edge devices to high-performance servers, meeting various scenario demands. In terms of model architecture, the traditional C3 module was replaced with the C2f module based on YOLOV8,coupled with the SPPF module Error! Reference source not found., achieving model lightweighting while significantly enhancing detection performance. Experimental results reveal that this algorithm achieves a remarkable detection accuracy of 97.4% for luosifen packaging, with a recall rate of 97.6% and a mean Average Precision (mAP) of 97.0%, demonstrating exceptional detection precision and robust detection capabilities for the target. Error! Reference source not found.