Abstract: In order to achieve accurate and fast identification of citrus fruits, an improved YOLOv7 network model is proposed. The ELAN-CA structure was designed by combining the ELAN structure with the CA attention mechanism. The optimal combination point between the ELAN-CA structure and the YOLOv7 Backbone was studied, and the ELAN structure at the corresponding position was replaced to enhance the learning ability of the network. In order to further improve the recognition rate of the network, the CIOU loss function was replaced by the Focal-EIOU loss function, which solved the problem of loss function degradation. The model's average recognition accuracy was improved from 96.91% to 97.56%. Experimental results show that the CA attention mechanism and Focal-EIOU loss function can effectively improve the ability of YOLOv7 to identify citrus fruits in natural environments.