Abstract: The spinning workshop is an important part of the production of chemical fibers. The inspection of the silk road to find problems in time and deal with them back in time is essential for the stable production of the subsequent winding. The realization of round-the-clock and uninterrupted silk road inspection through robots mounted on the visual inspection system has become one of the core contents of the digital transformation of the spinning workshop. In industrial sites, it is necessary to take into account the detection beat and detection accuracy. Due to the small number of defective or abnormal samples in normal production, based on this, this paper uses a small sample target detection FSOD for the identification of fine silk threads to identify and cut abnormal areas; Fast SAM is used for target detection with obvious characteristics of guide wire hooks and nozzles. Recognition and cropping of abnormal positions; the cropped abnormal area is then recognized by YOLOv11, and the final recognized image is merged and displayed in the entire picture. Through the engineering application of the silk road inspection robot, the effectiveness of the algorithm is verified and the digital construction of chemical fiber enterprises is