Abstract: In response to low accuracy and slow speed of blood cell recognition in the natural state, this paper proposes a blood cell classification method of the residual network incorporating a multiple attention mechanism. In order to improve the computational speed of the network and enhance the nonlinear representation capability of the model, an attention blending unit module is proposed. In order to improve the representation capability of the model for blood cell features, a multiple attention mechanism is embedded. With an aim to further mitigate network overfitting and enhance the generalization capability of the model, the residual branch structure is optimized. The experimental results show that the model has an accuracy of 95.67% on the blood cell dataset with a parametric number of 13.22 M. In comparison with other networks, the model proposed in this paper has a higher accuracy while maintaining a lower parametric number.