Aiming at the characteristics of flexible cables in missile assembly—such as diverse models, complex colors, and high-reliability requirements—and to address the challenge of identifying and matching cables before automated welding, a cable label recognition and matching algorithm model was proposed based on the YOLOv8 neural network for object detection. A dataset was constructed for recognizing cable labels commonly used in various missiles. By employing Mosaic image augmentation, the diversity of the training dataset was enhanced, enabling accurate identification of cable labels under complex working conditions. Furthermore, a color recognition algorithm based on label position was developed, achieving precise extraction and matching of both the content and color information of cable labels. Experimental results demonstrate that, with the introduction of prior knowledge and a re-identification mechanism, the recognition accuracy of the algorithm approaches 100%. In terms of recognition efficiency, while maintaining real-time detection capability, the average processing time per image is less than 30 milliseconds, significantly saving time without compromising reliability. The deep learning-driven cable identification and matching method developed can efficiently and reliably complete the automatic recognition and pairing of missile cables, providing key technical support for automated cable recognition and welding. This promotes the development of military assembly toward greater intelligence and high reliability, demonstrating promising application prospects.