Research on Pet Dog Species Identification Based on Convolution Neural Network

At present, the related research of image recognition is getting more and more popular, but in the process of research, the recognition effect of the model is not good enough and it is easy to misrecognize. This paper proposes an improvement solution for the above problems on the selection and construction of the model structure and the adjustment and optimization methods in the model training process. The final result achieves 96% recognition accuracy on the data composed of 9092 pet dog images. It is proved that the model by choosing deep-level network model and adopts regularization method to adjusting and optimizing the model, which can effectively improve model for image recognition effect.