Butterfly Recognition Based on Faster R-CNN

There are more than 18,000 butterflies in the world. Butterflies are essential cultural insects and have great significance in the field of insect research. However, due to its high similarity, various type, and the characteristics of the distinction is not apparent, the recognition and classification of butterflies at this stage has the problems of low accuracy and slow recognition speed, so it is of great significance to make research on the automatic identification of butterflies and improve the efficiency of automatic identification of butterflies. This paper uses the Faster R-CNN algorithm for butterfly recognition and describes the whole process from butterfly dataset selection, butterfly dataset processing, to butterfly classification. The experimental results show that the butterfly automatic recognition system based on the Faster R-CNN deep learning framework can realize automatic detection and species identification of butterfly photos in the ecological environment, and the average classification accuracy can reach 70.4%.