Study on Visual Convolutional Neural Network of Image Features of Mites
Mites are ancient micro-parasitic pests,which are closely related to people′s health.Using data-driven deep learning technology to conduct feature learning analysis on large-scale mite images will greatly pro-mote the research progress of related disciplines.This paper introduced artificial modeling,classical convolutional neural network algorithms such as AlexNet and VGG16,and combined with feature map visualization to explain the deep working mechanism of convolutions,which has certain research value for the biological fields related to mites and advanced computer vision methods.The experiment proves that the data feature extraction structure adopted in this paper is not only suitable for the visualization research of mite images,but also can be extended to other fields with certain robustness,in order to provide a universal method for visualization experiments and research of other disciplines.
NETL
NSTL
万方数据
