Object Detection and Machine Vision Classification of Traffic Markers Based on YOLOv8 Model

This paper discusses the research of object detection of traffic signs based on YOLOv8 model, aiming to evaluate the application potential of this model in the field of machine vision. After the training and testing of the model, we selected 30 EPOchs to carry out experiments, and the results showed that the YOLOv8 model could effectively identify and classify most traffic signs, showing good target detection performance. During the training process, by analyzing the confusion matrix, we observed that the model showed high accuracy on different types of traffic signs. This result shows that YOLOv8 can not only quickly locate the target, but also achieve accurate classification in multiple types of traffic signs. With the deepening of training, the model gradually learns more complex features, thus improving its detection ability. By monitoring the loss and mAP of the training set and validation set, we found that the loss value gradually decreased from the initial 0.9 to about 0.4, and finally stabilized. This trend reflects that the model gradually converges during the learning process, indicating that the optimization effect is remarkable. At the same time, both in the training set and the verification set, the mAP value also showed a continuous rise and stabilized at about 0.9, which further verified the efficiency and reliability of the model in the target detection task. In addition, by calculating F1 scores, Precision, Recall, and plotting accuracy-recall curves, we were able to more fully evaluate the performance of the YOLOv8 model. The results show that the model has excellent performance in traffic sign target detection and classification tasks, and all indexes have reached a satisfactory level. These data support our conclusion that YOLOv8 has strong potential for practical applications. To sum up, this study demonstrates the excellent ability of YOLOv8 model in traffic sign detection, which not only proves its adaptability to target recognition and classification tasks in complex scenes, but also lays a foundation for the future application of this technology in intelligent transportation system, automatic driving and other fields. In the future, we look forward to further exploring the application of this technology in a wider range of fields to drive the development and innovation of machine vision."