Tree Species Identification Based on the Fusion of Multiple Deep Learning Models Transfer Learning

The automatic identification of tree species images is of great value in practical application. The conventional identification algorithm based on hand-crafted features has a complex process of feature extraction, and it is difficult to make an adaptive optimal adjustment based on actual data and specific identification tasks, which is not suitable for the accurate identification of tree species images with a complex background. A tree species identification method based on the fusion of multiple deep learning models transfer learning is proposed to solve the problem of tree species images identification with a complex background in natural scenes. A new tree image dataset called TreesNet was built. Based on the TreesNet dataset, a variety of experiments were designed, and the method proposed was compared with the conventional image classification methods. The experimental results show that the image identification accuracy of the tree species in the complex background with the method proposed in this paper reaches 93.75%, which is much better than the conventional machine learning method.