Fine-Grained Bird Recognition Based On Deep Learning

China has a vast,wide range of animals,and birds are one of the most important species,but our country is lagging behind for the protection technologies of birds.Nowadays,in the context of the accumulation and application of big data,deep learning technology has developed rapidly and has achieved remarkable results in the field of computer vision.This paper proposed the application of deep learning methods to bird detection and identification analysis.It is of great significance for the realization of bird automatic identification technology,automatic bird monitoring technology,and the management of bird populations in specific regions.In this paper,we applied the deep learning method to image recognition technology.We proposed deep convolutional neural network to the research,and selected the Caltech-UCSD Birds-200-2011 database for fine-grained bird recognition.This paper has achieved three parts of fine-grained bird identification tasks:bird detection and identification,bird key point location,and bird segmentation.By analyzing the principles,we used two types of feature extraction networks(VGGNet,ResNet)to build a network structure based on a region proposal network for bird detection and identification tasks.Based on the results of detection and recognition network,we built the key point positioning sub-network,the network shared the features of the convolution layers,and finished the bird key point detection task.Finally,we did image segmentation task by fully convolution network.