“New breakthroughs have been made in the field of facial recognition research. In response to the problems of large parameter count, high computational resource consumption, and low recognition accuracy in traditional facial expression recognition methods, researchers have proposed a lightweight facial expression recognition method based on conditional coordinated attention mechanism. By reducing the number of layers in the MobileNet V3 network, increasing the number of channels, and using the Mish activation function, the nonlinearity after feature extraction has been achieved. Meanwhile, introducing an improved coordinated attention mechanism can capture detailed information of facial expressions in spatial and channel positions. The experimental results on the publicly available datasets FERPlus and RAF-DB show that the new method reduces the parameter count by 15.91% and achieves accuracy rates of 88.84% and 85.90%, respectively. Compared with the improved model, the accuracy rates have increased by 0.83% and 1.39%, respectively. This research achievement provides a new solution for the field of facial expression recognition and is expected to play an important role in practical applications.”