Method and device for classifying mirror-like images
The invention discloses a method and a device for classifying mirror-like images. The method comprises the following steps of marking a first object of a certain amount of mirror-like images, inputting the marked mirror-like images into a deep learning detection model, training through a convolution...
Saved in:
Main Authors | , , |
---|---|
Format | Patent |
Language | Chinese English |
Published |
17.10.2017
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | The invention discloses a method and a device for classifying mirror-like images. The method comprises the following steps of marking a first object of a certain amount of mirror-like images, inputting the marked mirror-like images into a deep learning detection model, training through a convolutional neural network, and learning the characteristics of the first object; after the training process is converged, obtaining a deep learning detection model, and accurately locating the position of the first object in a newly input mirror-like image; according to the attribute of the position of the first object in the newly input mirror-like image, classifying a target image. Compared with the manual classification method, the above method is efficient and accurate. The distortion of an analysis result can be guaranteed to be very small.
本发明公开了种对类镜像图像分类的方法及装置,其中,该方法包括:对定量的类镜像图像的第对象进行标注,并将标注好的类镜像图像输入深度学习检测模型,训练卷积神经网络,学习第对象的特征,当训练过程收敛后,得到的深度学习检测模型将具备能够准确定位出第目标在输入新的类镜像照片中的所在位置。再根据第对象在类镜像图像中的位置的属性将目标图像分类,相比人工分类,高效、准确。可 |
---|---|
Bibliography: | Application Number: CN201710387498 |