Abstract
In recent years, Deep Learning is widely used in various fields, such as object detection, speech recognition, etc. This paper focuses on the application of convolutional neural networks in transfer leaning. Convolutional Neural Networks (CNN) is a branch of Deep Learning. The calculation speed of matrix and CNN's training speed have been improved greatly for GPU's excellent computing capability. In the process of practice, the CNN model shows strong generalization ability. The CNN model can achieve higher classification accuracy with fine-tuning adjustment under the premise of sacrificing certain performance.