Trialling a convolution neural network for the identification of Braconidae in New Zealand

Computer vision approaches, such as deep learning, potentially offer a range of benefits to entomology, particularly for the image-based identification taxa. An experiment was conducted gauge ability convolution neural network (CNN) identify genera Braconidae from images forewings. A learning CNN trained via transfer small set 488 57 genera. Three-fold cross-validation achieved an accuracy 96.7%, thus demonstrating that genus using forewings is highly predictive. Further work needed increase both coverage species level and number available.