Data for: Fine-Grained Visual Categorization of Butterfly Specimens at Sub-species Level Via a Convolutional Neural Network with skip-connections

Published: 31 March 2020| Version 1 | DOI: 10.17632/s8vzkb8t2r.1
Wanlin Gao,


For performance evaluation, a total of 24,836 images of butterfly specimen spanning 56 sub-species were acquired as benchmark dataset for their strong similarity with subordinate categories. The camera used is Canon EOS 5D Mark IV and the shooting distance was three to seven cm depending on the worm size. The image format was JPEG and each one was a 24-bit color bitmap. Each image was classified into one corresponding ground truth category with the help of entomology experts. It is an interesting but challenging dataset for performance verification of fine-grained visual categorization of butterfly specimens.



Artificial Neural Networks, Image Processing, Biological Classification, Image Classification, Deep Learning