基于改进YOLOv8n的通信终端识别算法OA

To address the issue of decreased identification accuracy of communication terminals caused by signal interference in complex electro-magnetic environments,an improved YOLOv8n-based emitter identification algorithm for communication terminals is proposed,named EMI-YO-LO.Firstly,to tackle the problem of interference signals occluding the target signal,a C2fCE module is proposed,which integrates deep convolu-tion,pointwise convolution,and the Efficient Channel Attention(ECA)mechanism to expand the model's receptive field.Secondly,a partial self-attention mechanism is embedded at the end of the backbone network to enhance the model's ability to learn signal features.Furthermore,five data augmentation strategies are employed to effectively expand the dataset.The experimental results indicate that EMI-YOLO demonstrates a 7.4%improvement in mAP50-95 than YOLOv8n in the training set,with a reduction of 0.4M in model parameters;compared to three base-line algorithms,EMI-YOLO improves the identification accuracy for six mobile phone models by an average of 42.3%,52%,53.4%,50.4%,34%and39.7%in the test set,respectively.Therefore,EMI-YOLO exhibits strong anti-interference capability and robustness in complex electromagnetic environments.