A Multi-Strategy Integrated Improved YOLOv8n Algorithm and its Application in Real-Time Target Detection

The detection task of targets in YOLOv8n often suffers from insufficient recognition accuracy and insufficient scene applicability, which makes it difficult to accurately recognize targets in complex scenes. To do this, we propose a multi-strategy integrated optimization YOLOv8n algorithm to meet the needs of high accuracy and high scene adaptability for autonomous driving. Specifically, to address the problem of low adaptability of C2f in scenes with rich features, the ODConv module is introduced to improve the adaptability of the model. In view of the problems of high computational cost and limited optimization capability of the CIOU loss function processing, the Focal_GIOU loss function is introduced to improve the accuracy of the model for target localisation, and also enhance its ability to detect targets in complex scenes. Finally, Nadam optimizer is applied to replace the SGD optimizer to improve the training efficiency of the model. The proposed algorithm is applied to the COCO128 dataset for validation, and the experimental results show that the improved YOLOv8n algorithm proposed in this paper has the advantage of high detection efficiency. The [email protected] of the model reaches 91.87%, the validity and applicability of the model are verified. Compared with other models, the growth advantage of [email protected] is up to 6.23%, which proves the superiority of the model. The improved model proposed in this paper is used to solve the problem of insufficiently accurate target detection and low operational efficiency in unmanned vehicles, and has a good application background.