Deep learning is one of the AI technologies that make accurate and efficient decisions. AI is able to perform multi-layered analysis with neural network and discover potential features. In this study, we examined a blood morphology analysis AI model with the deep learning method. The AI model learned with training images of mature white blood cells (WBC) that show typical morphology, and parameter tuning for optimization was performed. The AI model obtained by transfer learning calculated the classification prediction value with the test image. And it compared with visual classification by clinical laboratory technologists. Two classifications between the specific cell group and the mixed cell group showed an accuracy of 82.6 to 100%. After pre-training with background-removal images, addition-al transfer learning and fine-tuning for six classifications was performed on original images with background cells. The learning model showed 99% accuracy, and a highly accurate AI model for leukocyte classification was obtained. It is considered that incorrect classification can be detected and corrected by performing two-classification analysis on the results obtained by six-classification analysis. This AI model is useful as a leukocyte classification and screening technique in cell groups showing a typical cell image.