Effective classification of birds’ species based on transfer learning

In recent years, with the deterioration of earth’s ecological environment, survival birds has been more threatened. To protect and diversity species on earth, it is urgent to build an automatic bird image recognition system. Therefore, this paper assesses performance traditional machine learning deep models recognition. Also, help-ability transfer in field tested evaluate best model for systems. Three groups classifiers were constructed, namely, based algorithms, convolutional neural networks, learning-based networks. After experiments, these three showed significant differences classification effect Kaggle-180-birds dataset. The experimental results finally prove that effective than algorithms as number increases. Besides, obtained show when sample data small, can help network classifier improve accuracy.