Abstract: In underwater optical image target detection tasks, complex underwater environments, light attenuation, and the prevalence of small underwater targets significantly affect detection accuracy. To improve this accuracy, an underwater target detection algorithm based on YOLOv5s is proposed. Firstly, the Spd-Conv module is introduced into the backbone network to enhance the recognition capability of small targets, effectively improving the detection accuracy. Secondly, the YOLOx_head module is added to the prediction network to enhance the model's convergence speed by using decoupled detection heads, enabling rapid convergence during training. Finally, building upon the existing Mixed Local Channel Attention mechanism, the C3-MLCA module is designed to capture local spatial information, thereby enhancing the model's ability to capture target features and further improving detection accuracy. Experimental results demonstrate that the improved algorithm increases a 2.0% increase mAP@0.5 by 2.0%, reaching 84.1%, and improves mAP@0.5:0.95 by 3.0%, reaching 48.0%. These results confirm the effectiveness of the proposed algorithm and the enhancement in detection accuracy.