A Multi-Strategy Integrated Improved Yolov8n Algorithm and Its Application to Automatic Driving Detection

For the existing target detection algorithms with low detection accuracy and slow recognition efficiency, it is difficult to meet the needs of autonomous driving to enable vehicles to quickly and accurately identify and localize the surrounding objects. For this reason, this paper proposes a multi-strategy integrated and improved Yolov8n algorithm to meet the demand for fast and accurate identification and localisation of surrounding objects for autonomous driving. Firstly, for the problem of low detection accuracy of Yolov8n, the GAM attention mechanism is introduced to improve the model's ability to capture key information. Subsequently, for the problem of poor model robustness, the CIOU loss function is replaced with the GIOU loss function to improve the model robustness. Finally, the SGD optimizer is applied to optimize the hyperparameters of the model to improve the model training efficiency. The proposed improved algorithm is applied to the VOC2007 dataset for validation, and the experimental results show that the improved model on the validation set [email protected] can be 86.5%. The effectiveness and applicability of the proposed algorithm is verified by the improved model in the validation set. The Precision, Recall and [email protected] all have significantly improved. The proposed improved Yolov8n model was compared with other target detection models, [email protected] improved 2.8%~12.3%, and the superiority of the model is verified.